CN111191648B

CN111191648B - Method and device for image recognition based on deep learning network

Info

Publication number: CN111191648B
Application number: CN201911389478.7A
Authority: CN
Inventors: 陆舟; 于华章
Original assignee: Feitian Technologies Co Ltd
Current assignee: Feitian Technologies Co Ltd
Priority date: 2019-12-30
Filing date: 2019-12-30
Publication date: 2023-07-14
Anticipated expiration: 2039-12-30
Also published as: CN111191648A

Abstract

The invention discloses a method and a device for image recognition based on a deep learning network, and relates to the field of image recognition. The device transmits the first parameter data to a deep learning network, and identifies all pictures acquired from a picture sample library to obtain a second list; sequentially carrying out fifth operation on the predicted frame data in the second list to obtain a fifth list: obtaining total deviation data according to all deviation data in the fifth list, comparing the total deviation data with preset deviation data, and when the total deviation data is larger than or equal to the preset deviation data, updating the first parameter data by using a fourth preset algorithm according to the total deviation data and the preset deviation data, and emptying the second list and the fifth list; and when the total deviation data is smaller than the preset deviation data, processing the picture to be identified by using the deep learning network containing the first parameter data to obtain a prediction frame data table, and ending.

Description

Method and device for image recognition based on deep learning network

Technical Field

The invention relates to the field of image recognition, in particular to a method and a device for image recognition based on a deep learning network.

Background

Deep Learning (DL, one of the key research areas of machine Learning technology in artificial intelligence) is to implement artificial intelligence in a computing system by establishing an artificial neural network (Artifitial Neural Networks, ans) having a hierarchical structure. The hierarchical artificial neural network used by the method has various forms, and the complexity of the hierarchy is generally called depth; deep learning uses data to update parameters in its construction to achieve training goals, a process known as "learning" that can be used to learn high-dimensional data for complex structures and large samples;

however, existing image recognition methods are typically: manually designing an algorithm for an object on a picture to be identified; then, the pictures containing the object images are transmitted into an algorithm one by one, and parameters of the algorithm are manually and gradually adjusted according to the deviation condition of the images on the pictures identified by the algorithm until the deviation of the images meets the design requirement; the designed algorithm is only applicable to objects on pictures used in the algorithm design process, cannot be applicable to the identification of images of other objects, and if other objects are to be identified, a new algorithm must be designed for the other objects to be identified again, so that the model generalization performance is poor; the parameters in the algorithm are adjusted repeatedly by manpower according to the deviation condition of the identified images on each picture, and the parameter adjusting method is high in manpower dependence and time-consuming and labor-consuming.

With the development of social economy, huge image information is generated every day; the application of deep learning networks suitable for large sample data to the field of image recognition to identify images has become a most urgent need to solve the above problems.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provides a method and a device for image recognition based on a deep learning network.

The invention provides a method for image recognition based on a deep learning network, which comprises the following steps:

step S0: the device transmits the first parameter data to a deep learning network, and acquires a picture sample library from a picture database;

step S1: when a picture is obtained from the picture sample library, the device divides the picture into first preset data grids according to the first preset data and calculates grid coordinate data of each grid; obtaining a prediction frame data table by using a deep learning network comprising first parameter data to identify the acquired picture according to the first preset data, the second preset data and the third preset data, wherein the prediction frame data table comprises second preset data which correspond to all grids respectively; sequentially acquiring grids from a first preset data grid and grid coordinate data corresponding to the grids, and executing a first operation on the acquired grids;

The first operation is: the device determines the image corresponding to the grid and the actual coordinate data of the image corresponding to the grid according to the grid coordinate data and the image actual coordinate data table corresponding to the picture; sequentially obtaining prediction frame data from second preset data corresponding to the grid, and executing a second operation on the obtained prediction frame data: the second operation includes steps S1-01 to S1-03:

step S1-01: the device acquires all category probability data from the prediction frame data, selects the largest category probability data from all category probability data, and determines the category data of the prediction frame data according to the largest category probability data;

step S1-02: the device acquires an image corresponding to the grid and actual coordinate data of the image; sequentially acquiring an image and actual coordinate data of the image from the image corresponding to the grid and the actual coordinate data of the image, and executing a third operation on the acquired image and the actual coordinate data of the image; sequentially obtaining the images from the image table corresponding to the same category of the prediction frame, and executing a fourth operation on the obtained images:

The third operation is: the device acquires category data of the image from the actual coordinate data of the image, and correspondingly stores the prediction frame data, the image and the actual data of the image into a prediction frame corresponding same-category image table when the acquired category data of the image is identical with the category data of the prediction frame data;

the fourth operation is: when the device acquires the actual coordinate data of the image from the corresponding image table of the same category, calculating the predicted frame data and the actual coordinate data to obtain a second confidence parameter; calculating the confidence coefficient corresponding to the image according to the first confidence parameter, eleventh preset data, twelfth preset data and second confidence parameter in the predicted frame data, and correspondingly storing the confidence coefficient, the predicted frame data, the category data and the actual coordinate data of the image into a first list;

step S1-03: the device obtains the maximum confidence in the first list; judging whether the maximum confidence coefficient is larger than the first comparison data, correspondingly storing the prediction frame data corresponding to the maximum confidence coefficient and the actual coordinate data of the image obtained from the first list to the second list, and emptying the first list and the image table of the same category corresponding to the prediction frame; otherwise, the first list and the corresponding image table of the same category of the prediction frame are emptied;

Step S2: when all pictures in the picture sample library are acquired, the device sequentially acquires prediction frame data from the second list, and executes a fifth operation on the acquired prediction frame data;

the fifth operation is: the device calculates deviation data of the predicted frame data and the actual coordinate data according to the predicted frame data and the actual coordinate data, and stores the deviation data into a fifth list;

step S3: the device obtains total deviation data according to all deviation data in the fifth list, compares the total deviation data with preset deviation data, and executes step S4 when the total deviation data is greater than or equal to the preset deviation data; when the total deviation data is smaller than the preset deviation data, processing the picture to be identified by using a deep learning network containing first parameter data to obtain a prediction frame data table, and ending;

step S4: the device updates the first parameter data by using a fourth preset algorithm according to the total deviation data and the preset deviation data, clears the second list and the fifth list, and returns to the step S0.

The invention also provides a device for carrying out image recognition based on the deep learning network, which comprises: the device comprises an incoming module, an acquisition module, an execution module, a deviation comparison module, a module to be identified and a parameter updating module;

The input module is used for transmitting the first parameter data to the deep learning network;

the acquisition module is used for acquiring a picture sample library from the picture database;

the acquisition module is also used for acquiring a picture from the picture sample library, dividing the picture into first preset data grids according to the first preset data and calculating grid coordinate data of each grid; obtaining a prediction frame data table by using a deep learning network comprising first parameter data to identify the acquired picture according to the first preset data, the second preset data and the third preset data, wherein the prediction frame data table comprises second preset data which correspond to all grids respectively; sequentially acquiring grids from a first preset data grid and grid coordinate data of the grids;

the execution module is configured to execute a first operation on the grid acquired by the acquisition module:

the execution module comprises a first execution operation unit;

the first execution operation unit comprises a determination module unit and a second execution operation unit;

the determining module unit is used for determining the image corresponding to the grid and the actual coordinate data of the image corresponding to the grid according to the grid coordinate data of the grid and the image actual coordinate data table corresponding to the picture, which are acquired by the acquiring module; acquiring prediction frame data from second preset data corresponding to the grid in sequence, and triggering a second execution operation unit;

The second execution operation unit is configured to perform a second operation on the obtained prediction frame data of the determination module unit:

the second execution operation unit comprises a first determination unit, a third execution operation unit, a second acquisition unit, the fourth execution operation unit and a judgment, preservation and emptying unit;

the first determining unit is configured to obtain all category probability data from the obtained prediction frame data of the determining module unit, select the largest category probability data from all category probability data, and determine the category data of the prediction frame data according to the largest category probability data; acquiring an image corresponding to the grid and actual coordinate data of the image; acquiring an image and actual coordinate data of the image from the image corresponding to the grid and the actual coordinate data of the image in sequence, and triggering the third execution operation unit;

the third execution operation unit is configured to execute a third operation on the image acquired by the first determination unit and actual coordinate data of the image:

the third execution operation unit is configured to obtain category data of the image from the actual coordinate data of the image obtained by the first determination unit, and correspondingly store the prediction frame data and the image and the actual data of the image into a prediction frame corresponding same category image table when the obtained category data of the image is the same as the category data of the prediction frame data;

The second obtaining unit is used for obtaining images from the image table of the same category corresponding to the prediction frame obtained by the third execution operation unit in sequence;

the fourth execution operation unit is used for calculating the prediction frame data determined by the determination module unit and the actual coordinate data of the image acquired by the second acquisition unit to obtain a second confidence parameter; calculating the confidence coefficient corresponding to the image according to the first confidence parameter, eleventh preset data, twelfth preset data and second confidence parameter in the predicted frame data, and correspondingly storing the confidence coefficient, the predicted frame data, the category data and the actual coordinate data of the image into a first list;

the judging, storing and emptying unit is used for acquiring the maximum confidence coefficient in the first list obtained by the fourth execution operation unit; judging whether the maximum confidence coefficient is larger than the first comparison data, correspondingly storing the prediction frame data corresponding to the maximum confidence coefficient and the actual coordinate data of the image obtained from the first list to the second list, and emptying the first list and the image table of the same category corresponding to the prediction frame; otherwise, the first list and the corresponding image table of the same category of the prediction frame are emptied;

The obtaining module is further configured to obtain, when all the pictures in the picture sample library are obtained, prediction frame data and actual coordinate data of the corresponding image from the second list, and trigger the fifth execution operation unit;

the fifth execution operation unit is used for calculating deviation data of the predicted frame data and the actual coordinate data according to the predicted frame data and the actual coordinate data, and storing the deviation data into a fifth list;

the deviation comparison module is used for obtaining total deviation data according to all deviation data in a fifth list obtained by the fifth execution operation unit and comparing the total deviation data with preset deviation data;

the processing to-be-identified module is used for processing the to-be-identified picture by using the deep learning network containing the first parameter data to obtain a prediction frame data table when the deviation comparison module obtains that the total deviation data is smaller than the preset deviation data, and ending;

and the parameter updating module is used for updating the first parameter data by using a fourth preset algorithm according to the total deviation data and the preset deviation data when the deviation comparison module obtains that the total deviation data is smaller than the preset deviation data, and clearing the second list and the fifth list.

Compared with the prior art, the invention has the following advantages: the invention provides a method and a device for image recognition based on a deep learning network; the device automatically simulates and updates the first parameter data in the deep learning network until the deep learning network is successfully optimized; the images on the pictures can be comprehensively, stably, accurately, quickly and in real time identified through the optimized and successful deep learning network; determining the type of the image on the picture and the coordinate data of the image on the picture; images with different scales and sizes in the same category on the picture can be well identified; the method can be applied to the situation of large-scale image recognition on the picture, liberates manual labor force and improves the efficiency of recognizing the image on the picture.

Drawings

Fig. 1 is a flowchart of a method for performing image recognition based on a deep learning network according to a first embodiment of the present invention;

fig. 2-1 and fig. 2-2 are flowcharts of a method for performing image recognition based on a deep learning network according to a second embodiment of the present invention;

fig. 3 is a flowchart of a method for image recognition based on a deep learning network according to a third embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

Example 1

An embodiment of the present invention provides a method for image recognition based on a deep learning network, as shown in fig. 1, including the following steps:

step 100: the device transmits the first parameter data to a deep learning network, and acquires a picture sample library from a picture database;

step 101: when a picture is obtained from the picture sample library, the device divides the picture into first preset data grids according to the first preset data and calculates grid coordinate data of each grid; obtaining a predicted frame data table by using a deep learning network comprising first parameter data to identify the acquired picture according to the first preset data, the second preset data and the third preset data, wherein the predicted frame data table comprises second preset data which respectively correspond to all grids; sequentially acquiring grids and grid coordinate data corresponding to the grids from a first preset data grid, and executing a first operation on the acquired grids:

the first operation is as follows: the device determines the actual coordinate data of the image corresponding to the grid according to the grid coordinate data and the image actual coordinate data table corresponding to the picture; sequentially acquiring prediction frame data from second preset data corresponding to the grid, and executing a second operation on the acquired prediction frame data: the second operation includes steps 101-01 to 101-03:

Step 101-01: the device acquires all the category probability data from the prediction frame data, selects the maximum category probability data from all the category probability data, and determines the category data of the prediction frame data according to the maximum category probability data;

step 101-02: the device acquires an image corresponding to the grid and actual coordinate data of the image; sequentially acquiring the image and the actual coordinate data of the image from the image and the actual coordinate data of the image corresponding to the grid, and executing a third operation on the acquired image and the actual coordinate data of the image: sequentially acquiring images from the image table corresponding to the same category of the prediction frame, and executing a fourth operation on the acquired images:

the third operation is as follows: the device acquires category data of the image from the actual coordinate data of the image, and correspondingly stores the prediction frame data and the actual data of the image and the image into a corresponding category image table of the prediction frame when the acquired category data of the image is the same as the category data of the prediction frame data;

Step 101-03: the device obtains the maximum confidence in the first list; judging whether the maximum confidence coefficient is larger than the first comparison data, correspondingly storing the prediction frame data corresponding to the maximum confidence coefficient and the actual coordinate data of the image obtained from the first list to the second list, and emptying the first list and the image table of the same category corresponding to the prediction frame; otherwise, the first list and the corresponding image table of the same category of the prediction frame are emptied;

step 102: when all the pictures in the picture sample library are acquired, the device sequentially acquires the predicted frame data from the second list, and executes a fifth operation on the acquired predicted frame data:

the fifth operation is as follows: the device calculates deviation data of the predicted frame data and the actual coordinate data according to the predicted frame data and the actual coordinate data, and stores the deviation data into a fifth list;

optionally, the fifth operation includes the steps of: the device calculates deviation data of the predicted frame data and the actual coordinate data according to the predicted frame data and the actual coordinate data, and stores the deviation data into a fifth list.

Step 103: the device obtains total deviation data according to all the deviation data in the fifth list, compares the total deviation data with the preset deviation data, and executes step 104 when the total deviation data is greater than or equal to the preset deviation data; when the total deviation data is smaller than the preset deviation data, processing the picture to be identified by using a deep learning network containing first parameter data to obtain a prediction frame data table, and ending;

Step 104: the device updates the first parameter data using a fourth preset algorithm based on the total deviation data and the preset deviation data, clears the second list and the fifth list, and returns to step 100.

Optionally, the grid data includes a grid abscissa, a grid ordinate, a grid length, and a grid width;

correspondingly, the prediction frame data table comprises fourth data, and the prediction frame data comprises a prediction frame abscissa, a prediction frame ordinate, a prediction frame length, a prediction frame width, a first confidence parameter and category probability data of third preset data;

correspondingly, the fourth data has a binding relationship with the first preset data and the second preset data, and the binding relationship is as follows: the fourth data is the product of the first preset data and the second preset data;

further correspondingly, the first preset data is in the form of data S.

Optionally, the third operation includes the following steps D01 to D02:

step D01: the device acquires category data of the image from actual coordinate data of the image;

step D02: the device judges whether the category data of the image is the same as the category data of the prediction frame data, if so, the device stores the actual coordinate data of the image and the prediction frame data into a prediction frame same category image table; otherwise, ending.

Optionally, the fourth operation includes the following steps D11 to D12;

step D11: the device obtains the intersection area and the union area of the predicted frame data and the image according to the predicted frame data and the actual data of the image by an eleventh preset algorithm; calculating the intersection area and the union area to obtain a second confidence parameter;

further, in step D11, the apparatus obtains an intersection area and a union area of the prediction frame data and the image according to the prediction frame data and the actual data of the image, specifically: and calculating the abscissa, the ordinate, the length and the width of the predicted frame and the actual abscissa, the actual ordinate, the actual length and the actual width in the actual data of the image according to an eleventh preset algorithm to obtain the intersection area and the union area of the predicted frame corresponding to the first current predicted frame data and the second current image.

Further, in step D11, the intersection area and the union area are calculated to obtain a second confidence parameter, which specifically is: the device calculates the intersection area and the union area according to a first preset algorithm to obtain a second confidence parameter.

Step D12: the device acquires a first confidence parameter in the predicted frame data, and determines the confidence coefficient of the predicted frame data according to the first confidence parameter, the second confidence parameter, the fifth preset data and the sixth preset data; correspondingly storing confidence level of the predicted frame data, the predicted frame data and category data of the predicted frame data and actual data of the image into a first list;

Further, step D12 specifically includes: the device acquires a first confidence parameter in the predicted frame data, determines the category of the first confidence parameter, determines the confidence coefficient of the predicted frame data according to the fifth preset data and the second confidence parameter when the first confidence parameter is the fifth preset data, and correspondingly stores the confidence coefficient of the predicted frame data, the category data and the actual data confidence coefficient of the image into a first list; when the first confidence parameter is sixth preset data, determining the confidence coefficient of the predicted frame data according to the second confidence parameter, and correspondingly storing the confidence coefficient of the predicted frame data, the category data and the actual data of the image into a first list;

further, step D12 is specifically: the device acquires a first confidence parameter in the predicted frame data, determines the category of the first confidence parameter, and when the first confidence parameter is fifth preset data, calculates the fifth preset data and the second confidence parameter according to a second preset algorithm to obtain the confidence coefficient of the predicted frame data, and correspondingly stores the confidence coefficient of the predicted frame data, the category data and the actual data of the image into a first list to a first list; and when the first confidence parameter is sixth preset data, recording the second confidence parameter as the confidence coefficient of the predicted frame data, and correspondingly storing the confidence coefficient of the predicted frame data, the category data and the actual data of the image into a first list.

Further, in step D11, the intersection area and the union area are calculated to obtain a second confidence parameter, which specifically is: the device calculates the intersection area and the union area according to a first preset algorithm to obtain a second confidence parameter;

further, in step D11, the intersection area and the union area are calculated to obtain a second confidence parameter, which specifically is: the device carries out comparison operation on the intersection area and the union area according to a first preset algorithm to obtain a second confidence parameter;

further, in step 101-03, the predicted frame data and the actual coordinate data of the image corresponding to the maximum confidence coefficient obtained from the first list are correspondingly stored in the second list, specifically: obtaining prediction frame data corresponding to the maximum confidence coefficient and actual coordinate data of the image from the first list, and correspondingly storing the maximum confidence coefficient, the prediction frame data and the actual coordinate data of the image into the second list; the device acquires a fourth list from the second list;

the device acquires a fourth list from the second list, and comprises the following steps:

step D20: the device classifies the predicted frame data in the second list, and selects one category from the obtained categories as a first current category;

Step D21: the device acquires prediction frame data corresponding to the first current category, actual data of the image and confidence coefficient of the prediction frame data from the second list and stores the prediction frame data and the confidence coefficient of the prediction frame data into a third list;

step D22: the device acquires the maximum confidence coefficient from the third list, marks the predicted frame data corresponding to the maximum confidence coefficient, and marks the predicted frame data corresponding to the maximum confidence coefficient as first marked predicted frame data;

step D23: the device selects one predicted frame data except the first marked predicted frame data from the third list as second current predicted frame data;

step D24: the device acquires actual data corresponding to the second current prediction frame data from the third list; obtaining the intersection area and the union area of the prediction frames corresponding to the first marked prediction frame data and the prediction frames corresponding to the second current prediction frame data according to the prediction frame data corresponding to the first marked prediction frame data and the prediction frame data of the second current prediction frame; calculating the intersection area and the union area to obtain a third confidence parameter;

step D25: the device compares the third confidence parameter with the size of the third comparison data, and when the third confidence parameter is greater than or equal to the third comparison data, the step D26 is executed; when the third confidence parameter is smaller than the third comparison data, marking a second current prediction frame in a third list, and executing step D26;

Step D26: the device judges whether the predicted frame data which is not used as the second current predicted frame data exists in the predicted frame data except the marked predicted frame data in the third list, if so, the step D27 is executed; otherwise, correspondingly storing the marked prediction frame, the corresponding confidence coefficient, the actual data, the prediction frame data and the category data in the third list into a fourth list, emptying the third list, and executing the step D28;

step D27: the apparatus acquires next prediction frame data from the prediction frames other than the first marked prediction frame data in the third list as second current prediction frame data, and returns to step D24.

Step D28: the device judges whether the category which is not used as the first current category exists in the categories in the second list, if so, the step D29 is executed; otherwise, the second list is cleared, and a fifth operation is executed on each piece of prediction frame data in the fourth list;

step D29: the device selects the next category from the categories in the second list as the first current category, and returns to step D21.

Optionally, in step 103, processing the picture to be identified by using the deep learning network including the first parameter data to obtain a prediction frame data table includes the following steps:

Step T1: when a picture to be identified is acquired, the device divides the picture into first preset data grids according to the first preset data; identifying the picture to be identified by using a deep learning network comprising updated first parameter data according to the first preset data, the second preset data and the third preset data to obtain a prediction frame data table; the prediction frame data table comprises second preset data which correspond to all grids respectively and prediction frame data; sequentially acquiring grids from the prediction frame data table, and executing eleventh operation on the grids:

the eleventh operation is: acquiring prediction frame data from second preset data corresponding to the grid in sequence, and executing twelfth operation on the prediction frame data;

the twelfth operation is: the device acquires a first confidence parameter from the predicted frame data, and marks the predicted frame data in a predicted frame data table when the first confidence parameter is fifth preset data;

step T2: the device outputs all marked predicted frame data in the predicted frame data table;

further, in the twelfth operation, the predicted frame data is marked in the predicted frame data table, and replaced with: the device stores the predicted frame data to a twelfth list; sequentially obtaining prediction frame data from the twelfth list, and executing thirteenth operation on the prediction frame data;

The thirteenth operation is that when one piece of predicted frame data is obtained from the twelfth list, the device obtains all the category probability data from the predicted frame data, selects the largest category probability data from the obtained all the category probability data, determines the category data of the predicted frame data according to the largest category probability data, and correspondingly stores the category data and the predicted frame data in the twelfth list; classifying all prediction frame data in the twelve lists according to all the category data in the twelve lists; any category is acquired from the twelve lists, a fourteenth operation is executed, and when all the categories are acquired from the twelve lists, the twelfth list is emptied;

fourteen operations are: the device acquires the predicted frame data corresponding to the category from the twelfth list and stores the predicted frame data in the thirteenth list; acquiring the largest category probability data in all category probability data corresponding to the category from the prediction frame data in the thirteenth list; marking prediction frame data corresponding to the maximum category probability data in a thirteenth list; according to an eleventh preset algorithm, calculating the predicted frame data corresponding to the maximum category probability data and other predicted frame data in the thirteenth list respectively to obtain sixth confidence parameters corresponding to the other predicted frame data in the thirteenth list respectively; performing a fifteenth operation on each sixth confidence parameter; the fifteenth operation is: the device compares the sixth confidence parameter with the size of the sixth comparison data, and marks the predicted frame data in the thirteenth list when the sixth confidence parameter is smaller than the thirteenth confidence data; correspondingly marking all marked predicted frame data in the thirteenth list in a predicted frame data table; clearing the thirteenth list;

Further, according to the eleventh preset algorithm, the predicted frame data corresponding to the maximum class probability data and the other predicted frame data in the thirteenth list are respectively calculated to obtain sixth confidence parameters respectively corresponding to the other predicted frame data in the thirteenth list, which specifically are:

the device acquires the predicted frame data corresponding to the maximum category probability data from the thirteenth list and marks the predicted frame data as marked predicted frame data; when the predicted frame data except the marked predicted frame data is obtained from the thirteenth list, the device calculates the marked predicted frame data and the second current predicted frame data according to an eleventh preset algorithm to obtain an intersection area and a union area, and calculates the intersection area and the union area according to a first preset algorithm to obtain a sixth confidence parameter.

Example two

The second embodiment of the invention provides a method for image recognition based on a deep learning network; as shown in fig. 2-1 and 2-2, comprising the steps of:

before this embodiment, the following data are preset in the device by the user:

1. the picture sample library is a part of the picture database; in this embodiment, 10 optional pictures are selected from the picture database;

2. The actual data of the image on each picture, the actual data (the actual abscissa of the center point, the calibrated ordinate of the center point, the actual length, the actual width and the category data), wherein the actual abscissa, the actual length and the actual width in the actual data are all pixel coordinates;

3. wherein 3.1 is the first preset data (7X 7 grids, 7X7 in this embodiment); 3.2 is second preset data (2 in this embodiment) indicating the number of identifiable prediction frames in each grid; 3.3 is third preset data (2 in this embodiment) indicating the number of categories of images on pictures input into the deep learning network, wherein the first category is orifice and the second category is bdm;

4. first comparison data 0.142857143 and third comparison data 0.8;

5. presetting deviation data 3.5;

6. the first parameter data are parameter data which need to be transmitted into the deep learning network; the preset data transmitted by the user is at the beginning; the device updates the first parameter data according to the actual situation and then re-transmits the first parameter data to the deep learning network; for example, the first parameter data is matrix data including bias/weights;

Step 200: the device transmits the first parameter data to a deep learning network, and acquires a picture sample library from a picture database;

optionally, the method specifically comprises the following steps: the device randomly acquires a picture sample library from a picture database;

for example, the device randomly acquires 10 pictures from a picture database as a picture sample library;

step 201: the device acquires a picture from a picture sample library as a first current picture;

optionally, the method further comprises the following steps: the device preprocesses the sample pictures in the picture database, places the pictures generated by the preprocessing into the picture database of the sample pictures and stores the actual data of the images on the pictures generated by the preprocessing;

the pretreatment purposes in this step are two: firstly, increasing the sample capacity of the deep learning process; secondly, the images in the pictures are better identified; preprocessing includes smoothing and/or median filtering and/or geometric transformation and/or normalization and/or smoothing and/or restoration and/or enhancement and/or cropping and/or scaling and/or flipping and/or shifting and/or adding noise;

step 202: the device acquires actual data of all images in the first current picture and stores the actual data into an image actual coordinate data table;

For example, there are 8 images in the first current picture, and the actual data of all the images in the first current picture are shown in the following table 11:

375	279	16	16	orifice
					357	108	16	16	orifice
502.5	180	17	18	orifice
					279	363	18	18	orifice
458	322.5	18	17	orifice
					508.5	484.5	19	19	orifice
639	234	20	18	orifice
					768	211	18	18	orifice

step 203: dividing the first current picture into 7X7 grids of first preset data by the device to obtain a grid coordinate data table comprising 49 grid coordinate data of the first preset data, wherein the grid coordinate data comprises grid abscissas, grid ordinates, grid lengths and grid widths;

before this step, the user presets first preset data in the device; the first preset data represents the number of divided pictures into grids; for example, the first preset data is 7X7, i.e., the image is divided equally into 7 grid lengths in the transverse direction and 7 grid widths in the longitudinal direction; the user divides the first current picture into 7X7 grids, each grid corresponds to one piece of grid coordinate data, and one piece of grid coordinate data comprises a center coordinate (grid abscissa and grid ordinate), a grid length and a grid width;

for example, the data format of each grid coordinate data is: (grid abscissa of grid center, grid ordinate of grid center, grid length, grid width);

7X7 grid coordinate data on the first current picture are shown in Table 12 below:

/>

step 204: the device uses a deep learning network comprising first parameter data to identify a first current picture to obtain fourth data (7X 7X 2) prediction frame data, stores the fourth data (7X 7X 2) prediction frame data into a prediction frame data table, wherein the prediction frame data comprises 2 category probability data of a prediction frame abscissa, a prediction frame ordinate, a prediction frame length, a prediction frame width, a first confidence parameter and third preset data;

Optionally, each prediction frame data includes (4+1+third preset data 2) sub data, that is, a midpoint coordinate of the prediction frame, a prediction frame height and a prediction frame width of the prediction frame, and the target image in the prediction frame belongs to category probability data of each of C categories of the third preset data;

the prediction frame data is rectangular frame data containing each image recognized by the device; the abscissa of the prediction frame is the abscissa of the midpoint of the rectangular frame; the ordinate of the prediction frame is the ordinate of the midpoint of the rectangular frame; the length of the prediction frame is the length of the rectangular frame; the width of the prediction frame is the width of the rectangular frame;

before this embodiment, the user presets the second preset data 2 and the third preset data 2 in the device; the second preset data represents the number of identifiable prediction frames in each grid; the third preset data represents the total number of categories contained in the image on the picture input into the deep learning network; 4 represents four sub-data of the horizontal coordinate and the vertical coordinate of the prediction frame of the midpoint of the prediction frame, the length of the prediction frame and the width of the prediction frame, and all coordinates, lengths and widths on the image use pixel coordinates; 1 represents the sub-data of the first confidence parameter of each prediction block; the prediction frame data may be displayed as brackets (prediction frame abscissa of midpoint of the prediction frame, prediction frame ordinate of midpoint, prediction frame length, prediction frame width, first confidence parameter, third preset data, class probability data); the first confidence parameter indicates whether the target object falls into the grid corresponding to the prediction frame, if yes, the target object is 1, and if no, the target object is 0;

For example, in the present embodiment, the categories of the image include two of orifice and bdm, and the prediction frame data is as follows in fig. 1:

the first current picture obtains 7X7X2 prediction frame data and stores the 7X7X2 prediction frame data into a prediction frame data table, and the prediction frame data table is shown in the following chart:

/>

/>

/>

step 205: the device acquires one grid 5X6 in the first preset data 7X7 grids as a first current grid;

step 206: the device acquires 2 pieces of second preset data corresponding to the first current grid 5x6 from the prediction frame data table, and stores the 2 pieces of second preset data in the grid corresponding prediction frame table;

step 207: the device acquires one piece of prediction frame data from the grid corresponding prediction frame table as first current prediction frame data;

for example, the number of prediction frames corresponding to the grid is 2, and one prediction frame data (365, 280, 15, 18,1, 69.32850645, 30.67149355) is arbitrarily selected from the second preset data as the first current prediction frame data;

step 208: the device acquires first current prediction frame data from the grid corresponding prediction frame table; acquiring 2 category probability data of third preset data from the first current prediction frame data; selecting the largest category probability data from the 2 category probability data of the third preset data, and determining the category data of the first current prediction frame data according to the largest category probability data;

For example, the apparatus obtains first current prediction frame data from a prediction frame data table (365, 280, 15, 18,1, 69.32850645, 30.67149355); acquiring category probability data (69.32850645, 30.67149355) corresponding to the first current prediction frame data from the prediction frame data; acquiring 2 category probability data of third preset data corresponding to the first current prediction frame data from the prediction frame data; selecting the largest category probability data (69.32850645) from the third preset data 2 category probability data (69.32850645, 30.67149355), and acquiring category data orifice of the first current prediction frame data according to the largest category probability data (69.32850645);

step 209: the device determines the image and the actual data in the first current grid according to the image actual coordinate data table and the grid coordinate data of the first current grid; storing the image in the first current grid and the actual data of the image to a grid corresponding image table, and executing step 209-01;

step 209-01: the device acquires an image from the grid corresponding image table as a first current image;

step 209-02: the device acquires actual coordinate data of the first current image, and acquires category data of the first current image from the actual coordinate data of the first current image;

Step 209-03: the device judges whether the category data of the first current image is the same as the category data of the first current prediction frame data, if so, the device stores the actual coordinate data of the first current image and the first current prediction frame data into a prediction frame same category image table, and executes step 209-04, otherwise, executes step 209-04;

step 209-04: the device judges whether the next image serving as the first current image can be acquired from the grid corresponding image table, if yes, the device returns to step 209-05; otherwise, executing step 209-06;

step 209-05: the device acquires the next image from the grid corresponding image table as a first current image, and returns to step 209-02;

step 209-06: the device acquires an image from the image table of the same category of the prediction frame as a second current image, and executes step 210;

step 210: the device acquires actual data of a second current image from the image table with the same category as the prediction frame; acquiring an intersection area and a union area of the first current prediction frame data and the second current image according to the first current prediction frame data and the actual data of the second current image; calculating the intersection area and the union area to obtain a second confidence parameter;

optionally, the method specifically comprises the following steps: the device acquires actual data of the second current image from the image actual coordinate data table; calculating the abscissa, the ordinate, the length and the width of the predicted frame of the first current predicted frame data and the actual abscissa, the actual ordinate, the actual length and the actual width in the actual data of the second current image according to an eleventh preset algorithm to obtain the intersection area and the union area of the predicted frame corresponding to the first current predicted frame data and the second current image; calculating the intersection area and the union area by using a first preset algorithm to obtain a second confidence parameter;

Optionally, the step is more specifically: the device acquires actual data of the second current image from the image actual coordinate data table; calculating the abscissa, the ordinate, the length and the width of the predicted frame of the first current predicted frame data and the actual abscissa, the actual ordinate, the actual length and the actual width in the actual data of the second current image according to an eleventh preset algorithm to obtain the intersection area and the union area of the predicted frame corresponding to the first current predicted frame data and the second current image; performing a comparison operation on the intersection area and the union area to obtain a second confidence parameter;

for example, the device obtains actual data (375, 279, 16, orifice) of the second current image from an image actual coordinate data table; obtaining an intersection area 88 and a union area 438 of the first current prediction frame data and the second current picture from the first current prediction frame data (365, 280, 15, 18,1, 69.32850645, 30.67149355) and the actual data (375, 279, 16, orifice) of the second current picture; comparing the intersection area 88 with the union area 438 to obtain a second confidence parameter 0.200913242;

step 211: the device acquires a first confidence parameter in the first current prediction frame data, and determines the confidence coefficient of the first current prediction frame data according to the first confidence parameter, the second confidence parameter, the fifth preset data and the sixth preset data; correspondingly storing the confidence coefficient of the first current predicted frame data, the category data and the actual data of the second current image into a first list;

Optionally, the method specifically comprises the following steps: the device acquires a first confidence parameter in the first current predicted frame data, determines a category of the first confidence parameter, determines the confidence coefficient of the first current predicted frame data according to the fifth preset data and the second confidence parameter when the first confidence parameter is the fifth preset data, stores the confidence coefficient of the first current predicted frame data, the category and the actual data of the second current image into a first list, and executes step 212; when the first confidence parameter is the sixth preset data, determining the confidence coefficient of the first current prediction frame data according to the second confidence parameter, correspondingly storing the confidence coefficient of the first current prediction frame data, the prediction frame data and the actual data of the category and the second current image into a first list, and executing step 212;

optionally, the step is more specifically: the device acquires a first confidence parameter in the first current prediction frame data, determines the category of the first confidence parameter, calculates the fifth preset data and the second confidence parameter according to a second preset algorithm to obtain the confidence coefficient of the first current prediction frame data when the first confidence parameter is the fifth preset data, correspondingly stores the confidence coefficient of the first current prediction frame data, the category and the actual data of the second current image into a first list to a first list, and executes step 212; when the first confidence parameter is the sixth preset data, recording the second confidence parameter as the confidence coefficient of the first current prediction frame data, and corresponding the confidence coefficient of the first current prediction frame data, the prediction frame data and the category and the actual data of the second current image to the first list, and executing step 212;

For example, the apparatus obtains a first confidence parameter in the first current prediction frame data (365, 280, 15, 18,1, 69.32850645, 30.67149355), determines a category of the first confidence parameter, calculates a confidence 0.149980763498417 of the first current prediction frame data according to the second preset algorithm on the fifth preset data and the second confidence parameter 0.200913242 when the first confidence parameter is the fifth preset data 1, correspondingly stores the confidence 0.149980763498417 of the first current prediction frame data, the prediction frame data (365, 280, 15, 18,1, 69.32850645, 30.67149355), the category data orifice, and the actual data (375, 279, 16, orifice) of the second current image into a first list, and performs step 212; when the first confidence parameter is sixth preset data, recording the second confidence parameter as the confidence coefficient of the first current predicted frame data, correspondingly storing the confidence coefficient of the first current predicted frame data, the predicted frame data and the actual data of the category and the second current image into a first list, and executing step 212;

in this embodiment, the confidence level indicates the confidence level of the prediction frame in identifying the image;

step 212: the device judges whether the image table of the same category of the predicted frame has an image which is not used as the second current image, if yes, the step 213 is executed; otherwise, go to step 214;

Step 213: the device acquires the next image from the image table of the same category of the prediction frame as a second current image, and returns to the step 210;

step 214: the device empties the image table with the same category of the prediction frame, and acquires the maximum confidence coefficient from the first list and marks the maximum confidence coefficient as a first current confidence coefficient; comparing the first current confidence with the first comparison data, and acquiring actual data of the image corresponding to the first current confidence and category data of the first current prediction frame data from a first list when the first current confidence is larger than the first comparison data; correspondingly storing the first current confidence level, the actual data of the image, the first current prediction frame data and the category data of the first current prediction frame data in the second list, and executing step 215; when the confidence level of the first current prediction frame data is less than or equal to the first comparison data, executing step 215;

for example, the device obtains the greatest confidence level from the first list as the first current confidence level; comparing the first current confidence with the first comparison data 0.142857142857143, and when the first current confidence is greater than the first comparison data, acquiring actual data of the image corresponding to the first current confidence, first current prediction frame data and category data of the first current prediction frame data from the first list; correspondingly storing the first current confidence level, the actual data of the image, the first current prediction frame data and the category data of the first current prediction frame data in the second list, and executing step 215; when the confidence level of the first current prediction frame data is less than or equal to the first comparison data, executing step 215;

Step 215: the device clears the first list; judging whether the 2 predicted frames corresponding to the first current grid have predicted frames which are not used as the data of the first current predicted frame, if so, executing step 216; otherwise, go to step 217;

step 216: the device acquires the next predicted frame from the 2 predicted frames corresponding to the first current grid as the data of the first current predicted frame, and returns to the step 208;

step 217: the device classifies the predicted frame data in the second list according to the category data in the second list; selecting one category as a first current category;

for example, 8 images on the first current picture are selected to be orifices, so that the grid corresponding prediction frame table has only one category orifice;

step 218: the device acquires prediction frame data corresponding to the first current category, actual data of the image and confidence coefficient of the prediction frame data from the second list and stores the prediction frame data and the confidence coefficient of the prediction frame data into a third list; obtaining the maximum confidence coefficient from the third list, marking the predicted frame data corresponding to the maximum confidence coefficient, and marking the predicted frame data corresponding to the maximum confidence coefficient as marked predicted frame data;

step 219: the device selects one prediction frame data except the marked prediction frame data from the third list as second current prediction frame data;

Step 220: the device acquires actual data corresponding to the second current prediction frame data from the third list; obtaining the intersection area and the union area of the predicted frame corresponding to the marked predicted frame data and the predicted frame corresponding to the second current predicted frame data according to the predicted frame data corresponding to the marked predicted frame data and the predicted frame data of the second current predicted frame; calculating the intersection area and the union area to obtain a third confidence parameter;

step 221: the device compares the third confidence parameter with the size of the third comparison data, and when the third confidence parameter is greater than or equal to the third comparison data, step 222 is performed; when the third confidence parameter is less than the third comparison data, marking a second current prediction frame in a third list, and executing step 222;

for example, the device compares the third confidence parameter to the magnitude of the third comparison data of 0.8, and when the third confidence parameter is greater than or equal to the third comparison data, performs step 222; when the third confidence parameter 0 is less than the third comparison data 0.8, marking the second current prediction frame in the third list, and executing step 222;

step 222: the apparatus judges whether or not there is predicted frame data which is not used as the second current predicted frame data in the predicted frame data other than the marked predicted frame data in the third list, and if so, performs step 223; otherwise, correspondingly storing the marked predicted frame data in the third list and the corresponding actual data of the image in the fourth list, emptying the third list, and executing step 224;

Step 223: the device acquires the next predicted frame data from the predicted frame data except the marked predicted frame data in the third list as the second current predicted frame data, and returns to the step 220;

step 224: the device determines whether there are any more categories in the second list that are not being used as the first current category, and if yes, step 225 is executed; otherwise, the second list is cleared, and step 226 is performed;

step 225: the device obtains the next category from the categories in the second list as the first current category, returning to step 218;

step 226: the device judges whether any grid which is not used as the first current grid exists in the 7x7 grids of the first preset data, if so, step 227 is executed; otherwise, go to step 228;

step 227: the device acquires the next grid from the 7x7 grids as a first current grid, and returns to the step 206;

step 228: the device determines whether there are any pictures in the picture sample library that are not used as the first current picture, if yes, step 229 is executed; otherwise, executing step 231;

step 229: the device acquires the next picture from the picture sample library as a first current picture, and returns to the step 202;

step 231: the device acquires one piece of predicted frame data from the fourth list as third current predicted frame data;

Step 232: the device acquires actual data of the image corresponding to the third current prediction frame data from the fourth list; calculating the third current predicted frame data and actual data corresponding to the third current predicted frame data according to a third preset algorithm to obtain deviation data of the third current predicted frame data; storing deviation data of the third current prediction frame data to a fifth list;

step 233: the device determines whether there is any predicted frame data in the fourth list that is not used as the third current predicted frame data, and if so, proceeds to step 234; otherwise, the fourth list is cleared, and step 235 is performed;

step 234: the device acquires the next predicted frame data from the fourth list as the third current predicted frame data, and returns to the step 232;

step 235: the device obtains total deviation data according to the deviation data in the fifth list, compares the total deviation data with the preset deviation data, clears the fifth list when the total deviation data is greater than or equal to the preset deviation data, and executes step 236; when the total deviation data is smaller than the preset deviation data, processing the picture to be identified by using a deep learning network containing first parameter data to obtain a prediction frame data table, and ending;

Step 236: the device updates the first parameter data by using a fourth preset algorithm according to the total deviation data and the preset deviation data, and returns to the step 200;

for example, the fourth preset algorithm is a back propagation method;

total loss value of 213.6

The preset deviation data is 3.0.

Example III

The third embodiment of the invention provides a method for image recognition based on a deep learning network; the deep learning network used in the present embodiment is the deep learning network that has been optimized successfully in the second embodiment; as shown in fig. 3, the method comprises the following steps:

step 301: the device receives a picture to be identified which is transmitted by a user;

step 302: the device uses a deep learning network comprising first parameter data to identify the picture to be identified to obtain 7x7x2 prediction frame data and respectively corresponding prediction frame data, and stores the prediction frame data and the prediction frame data thereof into a prediction frame data table; the prediction frame data comprises 2 category probability data of a prediction frame data abscissa and a prediction frame data ordinate of a midpoint of the prediction frame data, a prediction frame data length, a prediction frame data width, a first confidence parameter and third preset data;

in the second embodiment before this embodiment, the user presets the first preset data in the device; the first preset data represents the number of divisions of the image into grids; the first preset data is generally 7x7, namely the image is transversely equally divided into S grid lengths and longitudinally equally divided into S grid widths;

Optionally, each prediction frame data includes (4+1+c) sub-data, that is, a midpoint coordinate (midpoint abscissa and midpoint ordinate) of the prediction frame data, a prediction frame data height and a prediction frame data width of the prediction frame data, and the target image in the prediction frame data belongs to category probability data of each of the C categories; the user presets data B and data C in the device; data B represents the number of prediction frame data identifiable in each grid; data C represents the image category of the image on the picture input into the deep learning network; 4 represents four sub-data of midpoint coordinates of the predicted frame data, predicted frame data length of the predicted frame data and predicted frame data width, and all coordinates, lengths and widths on the image use pixel coordinates; 1 represents the sub-data of the first preset parameter of each prediction frame data; the prediction box data may be displayed as: (center abscissa, center ordinate, predicted frame data length, predicted frame data width, first confidence parameter, C category probability data);

for example, when C is 2, the prediction frame data is as follows in fig. 1:

step 303: the device acquires a grid from 7X7 grids of the first preset data as a first current grid;

Step 304: the device acquires 2 pieces of second preset data corresponding to the first current grid from the prediction frame data table, and stores the 2 pieces of second preset data corresponding to the first current grid into the grid corresponding prediction frame table;

step 305: the device acquires one prediction frame data from the grid corresponding prediction frame table as fifth current prediction frame data;

step 306: the device acquires a first confidence parameter from the fifth current prediction frame data, determines the category of the first confidence parameter, and when the first confidence parameter is fifth preset data, stores the fifth current prediction frame data to a twelfth list, and executes step 307; when the first confidence parameter is the sixth preset data, executing step 307;

step 307: the device judges whether the next predicted frame data serving as the fifth current predicted frame data can be obtained from the grid corresponding predicted frame table, if so, the next predicted frame data serving as the fifth current predicted frame data is obtained from the grid corresponding predicted frame table, and the step 306 is returned; otherwise, go to step 308;

step 308: the device acquires one piece of predicted frame data from the twelfth list as sixth current predicted frame data;

step 309: the device acquires 2 category probability data of third preset data from the sixth current prediction frame data from the twelfth list; selecting the maximum category probability data from the 2 category probability data of the third preset data, and determining category data of the sixth current prediction frame data according to the maximum category probability data; storing category data of the sixth current predicted frame data and the sixth current predicted frame data to a twelfth list correspondingly;

Step 310: the device judges whether the next predicted frame data as the sixth current predicted frame data can be obtained from the twelfth list, if so, the next predicted frame data is obtained from the twelfth list as the sixth current predicted frame data, and returns to step 309; otherwise, executing step 311;

step 311: the device classifies the predicted frame data in the twelfth list according to the category data in the twelfth list; selecting one category as a second current category;

step 312: the device acquires the predicted frame data corresponding to the second current category from the twelfth list and stores the predicted frame data in the thirteenth list;

step 313: the device acquires the maximum category probability data in all category probability data corresponding to the second current category from the prediction frame data in the thirteenth list; marking prediction frame data corresponding to the maximum category probability data in a thirteenth list;

step 314: the device marks the predicted frame data corresponding to the maximum category probability data in a tenth list as second marked predicted frame data;

step 315: the device selects one prediction frame data except the second marked prediction frame data from the thirteenth list as second current prediction frame data;

Step 316: the device calculates a sixth confidence parameter according to the eleventh preset algorithm on the second marked predicted frame data and the second current predicted frame data;

step 317: the device compares the sixth confidence parameter to the magnitude of the sixth comparison data, and when the sixth confidence parameter is greater than or equal to the sixth comparison data, performs step 318; marking the second current prediction frame data in a thirteenth list when the sixth confidence parameter is less than the thirteenth confidence data, performing step 318;

step 318: the apparatus judges whether or not there is predicted frame data which is not used as second current predicted frame data among the predicted frame data other than the marked predicted frame data in the thirteenth list, and if so, performs step 319; otherwise, the predicted frame data of all the marks in the thirteenth list are correspondingly marked in the predicted frame data table, and step 320 is executed;

step 319: the device acquires the next predicted frame data from the predicted frame data except the second marked predicted frame data in the thirteenth list as the second current predicted frame data, and returns to the step 316;

step 320: the device clears the thirteenth list; judging whether the category which is not used as the second current category exists in the categories in the twelfth list, if so, executing step 321; otherwise, go to step 322;

Step 321: the device obtains the next category from the categories in the twelfth list as the second current category, returning to step 312;

step 322: the device clears the twelfth list; judging whether the grids which are not used as the second current grid exist in the 7x7 grids of the first preset data, if yes, acquiring the next grid from the 7x7 grids of the first preset data as the second current grid, and returning to the step 304; otherwise, outputting the marked predicted frame data in the predicted frame data table, and ending.

Example IV

The fourth embodiment provides a device for image recognition based on a deep learning network, which comprises an incoming module, an acquisition module, an execution module, a deviation comparison module, a module to be recognized and a parameter updating module;

the acquisition module is also used for acquiring a picture from the picture sample library, dividing the picture into first preset data grids according to the first preset data and calculating grid coordinate data of each grid; obtaining a predicted frame data table by using a deep learning network comprising first parameter data to identify the acquired picture according to the first preset data, the second preset data and the third preset data, wherein the predicted frame data table comprises second preset data which respectively correspond to all grids; sequentially acquiring grid coordinate data of grids from the first preset data grids;

The execution module is used for executing a first operation on the acquired grid of the acquisition module:

the execution module comprises a first execution operation unit;

a second execution operation unit configured to perform a second operation on the acquired prediction frame data of the determination module unit:

the second execution operation unit comprises a first determination unit, a third execution operation unit, a second acquisition unit, a fourth execution operation unit and a judgment, preservation and emptying unit;

a first determining unit configured to acquire all category probability data from the acquired prediction frame data of the determining module unit, select the largest category probability data from all the category probability data, and determine category data of the prediction frame data according to the largest category probability data; acquiring an image corresponding to the grid and actual coordinate data of the image; sequentially acquiring images and actual coordinate data of the images from the images and the actual coordinate data of the images corresponding to the grids, and triggering a third execution operation unit;

A third execution operation unit configured to execute a third operation on the image and the actual coordinate data of the image acquired by the first determination unit:

a third execution operation unit, configured to obtain category data of the image from the actual coordinate data of the image obtained by the first determination unit, and when the category data of the obtained image is identical to the category data of the prediction frame data, correspondingly store the prediction frame data and the actual data of the image and the image to a corresponding category image table of the prediction frame;

the second acquisition unit is used for sequentially acquiring images from the corresponding image table of the same category of the prediction frame obtained by the third execution operation unit;

a fourth execution operation unit for calculating the prediction frame data determined by the determination module unit and the actual coordinate data of the image acquired by the second acquisition unit to obtain a second confidence parameter; calculating the confidence coefficient corresponding to the image according to the first confidence parameter, eleventh preset data, twelfth preset data and second confidence parameter in the predicted frame data, and correspondingly storing the confidence coefficient, the predicted frame data, the category data and the actual coordinate data of the image into a first list;

the judgment and storage emptying unit is used for acquiring the maximum confidence coefficient in the first list obtained by the fourth execution operation unit; judging whether the maximum confidence coefficient is larger than the first comparison data, correspondingly storing the prediction frame data corresponding to the maximum confidence coefficient and the actual coordinate data of the image obtained from the first list to the second list, and emptying the first list and the image table of the same category corresponding to the prediction frame; otherwise, the first list and the corresponding image table of the same category of the prediction frame are emptied;

The obtaining module is further configured to, when all the pictures in the picture sample library are obtained, obtain prediction frame data and actual coordinate data of the corresponding image from the second list, and trigger the fifth execution operation unit;

a fifth execution operation unit for calculating deviation data of the predicted frame data and the actual coordinate data according to the predicted frame data and the actual coordinate data, and storing the deviation data to a fifth list;

the processing module to be identified is used for obtaining a prediction frame data table by processing the picture to be identified by using the deep learning network containing the first parameter data when the deviation comparison module obtains that the total deviation data is smaller than the preset deviation data and the deep learning network is successfully optimized;

and the updating parameter module is used for updating the first parameter data by using a fourth preset algorithm according to the total deviation data and the preset deviation data when the deviation comparison module obtains that the total deviation data is smaller than the preset deviation data, and emptying the second list and the fifth list.

accordingly, the predicted frame data is rectangular frame data of a rectangular frame that frames the image recognized by the apparatus; the abscissa of the predicted frame is the abscissa of the midpoint of the rectangular frame; the ordinate of the prediction frame is the ordinate of the midpoint of the rectangular frame; the length of the prediction frame is the length of a rectangular frame; the width of the prediction frame is the width of a rectangular frame;

correspondingly, the fourth data has binding relation with the first preset data and the second preset data;

further, the binding relationship is: the fourth data is the product of the first preset data and the second preset data;

further correspondingly, the first preset data is the product of the data S and the data S.

Optionally, the save acquisition unit includes acquiring as a subunit

Correspondingly, the acquisition unit is used for acquiring one image from all the images corresponding to the grid as an image;

accordingly, an acquisition subunit for acquiring actual coordinate data of the image acquired as the subunit, and acquiring category data of the image from the actual coordinate data of the image acquired as the subunit;

Correspondingly, a first judging subunit, configured to judge whether the category data of the image obtained as the subunit is the same as the category data of the prediction frame data, and if yes, trigger the storing subunit;

correspondingly, a storage subunit, configured to store the actual coordinate data and the prediction frame data of the image used as the subunit into the prediction frame same-category image table, trigger the second judging subunit, and otherwise trigger the second judging subunit;

correspondingly, the second judging subunit is used for judging whether the next image can be obtained from all the images corresponding to the grid as an image, if so, the next image is obtained from all the images corresponding to the grid as an image, and the obtaining subunit is triggered; otherwise, acquiring images from the image table corresponding to the same category of the prediction frame in sequence, and triggering a fourth operation execution unit.

Optionally, the fourth execution operation unit includes a first calculation subunit and a first storage subunit;

correspondingly, the first calculating subunit is used for obtaining the intersection area and the union area of the predicted frame data and the image according to the predicted frame data and the actual data of the image acquired by the second acquiring unit according to an eleven preset algorithm; calculating the intersection area and the union area to obtain a second confidence parameter;

Correspondingly, the first storage subunit is used for acquiring a first confidence parameter in the predicted frame data, and determining the confidence coefficient of the predicted frame data according to the first confidence parameter, the second confidence parameter of the first calculation subunit, the fifth preset data and the sixth preset data; and correspondingly storing confidence level of the predicted frame data, the predicted frame data and category data of the predicted frame data and actual data of the image into a first list.

Further, the first storage subunit is specifically configured to obtain a first confidence parameter in the prediction frame data, determine a class of the first confidence parameter, and when the first confidence parameter is fifth preset data, determine a confidence level of the prediction frame data according to the fifth preset data and the second confidence parameter, and correspondingly store the confidence level of the prediction frame data, the class data and an actual data confidence level of the image to the first list; when the first confidence parameter is sixth preset data, determining the confidence coefficient of the predicted frame data according to the second confidence parameter, and correspondingly storing the confidence coefficient of the predicted frame data, the category data and the actual data of the image into a first list;

further, the first storing subunit is specifically configured to obtain a first confidence parameter in the predicted frame data, determine a class of the first confidence parameter, and when the first confidence parameter is fifth preset data, calculate the fifth preset data and the second confidence parameter according to a second preset algorithm to obtain a confidence level of the predicted frame data, and correspondingly store the confidence level of the predicted frame data, the class data, and actual data of the image into a first list to a first list; and when the first confidence parameter is sixth preset data, recording the second confidence parameter as the confidence coefficient of the predicted frame data, and correspondingly storing the confidence coefficient of the predicted frame data, the category data and the actual data of the image into a first list.

Further, when the first calculating subunit is configured to obtain, according to an eleven preset algorithm, an intersection area and a union area of the predicted frame data and the image from the predicted frame data and the actual data of the image acquired by the second acquiring unit, the first calculating subunit is configured to calculate, according to the eleven preset algorithm, a predicted frame abscissa, a predicted frame ordinate, a predicted frame length, and an actual abscissa, an actual ordinate, an actual length, and an actual width in the predicted frame data and the actual data of the image, to obtain an intersection area and a union area of the predicted frame corresponding to the first current predicted frame data and the second current image;

further, when the first calculating subunit is configured to calculate the intersection area and the union area to obtain the second confidence parameter, the first calculating subunit is specifically configured to calculate the intersection area and the union area according to a first preset algorithm to obtain the second confidence parameter;

further, when the first calculating subunit is configured to calculate the intersection area and the union area to obtain the second confidence parameter, the first calculating subunit is specifically configured to calculate the ratio of the intersection area to the union area to obtain the second confidence parameter.

Optionally, when the determining and saving unit is configured to correspondingly save the predicted frame data and the actual coordinate data of the image corresponding to the maximum confidence level from the first list to the second list, the determining and saving unit is specifically configured to correspondingly save the predicted frame data and the actual coordinate data of the image corresponding to the maximum confidence level from the first list to the second list;

correspondingly, the acquisition module further comprises a first acquisition unit, wherein the first acquisition unit is used for acquiring a fourth list from the second list;

correspondingly, the first acquisition unit comprises a first classification subunit, a first storage subunit, a first marked subunit, a comparison subunit, a first acquisition subunit and a second storage subunit;

correspondingly, the first classification subunit is configured to classify the prediction frame data in the second list, and sequentially select one category from the obtained categories as a first current category;

correspondingly, the first storing subunit is configured to obtain, from the second list, prediction frame data corresponding to the first current category obtained by the first classification subunit, actual data of the image, and confidence level of the prediction frame data, and store the prediction frame data in the third list;

Correspondingly, the first marking subunit is configured to obtain the maximum confidence coefficient from the third list, mark the prediction frame data corresponding to the maximum confidence coefficient, and mark the prediction frame data corresponding to the maximum confidence coefficient as first marked prediction frame data;

correspondingly, the first obtaining subunit is further configured to select, from the third list, one prediction frame data other than the first marked prediction frame data as second current prediction frame data;

correspondingly, a second storage subunit is configured to obtain actual data corresponding to the second current prediction frame data from the third list; obtaining the intersection area and the union area of the prediction frames corresponding to the first marked prediction frame data and the prediction frames corresponding to the second current prediction frame data according to the prediction frame data corresponding to the first marked prediction frame data and the prediction frame data of the second current prediction frame; calculating the intersection area and the union area to obtain a third confidence parameter;

correspondingly, the comparison subunit is used for comparing the third confidence parameter obtained by the confidence parameter storage subunit with the size of the third comparison data; marking a second current prediction frame in a third list when the third confidence parameter is less than the third comparison data;

Correspondingly, the first storage subunit is further configured to, when the first obtaining unit obtains that all the prediction frame data is selected from the third list except the first marked prediction frame data, store the marked prediction frame in the third list and the corresponding confidence, actual data, prediction frame data, and category data in the fourth list, and empty the third list.

Optionally, the fifth execution operation unit is configured to calculate deviation data of the predicted frame data and the actual coordinate data according to the predicted frame data and the actual coordinate data, and store the deviation data in the fifth list.

Optionally, the processing module to be identified includes an eleventh acquisition unit, an eleventh execution unit and an eleventh output unit;

accordingly, an eleventh acquisition unit for dividing the picture into first preset data grids according to the first preset data; identifying the picture to be identified by using a deep learning network comprising updated first parameter data according to the first preset data, the second preset data and the third preset data to obtain a prediction frame data table; the prediction frame data table comprises second preset data which correspond to all grids respectively and prediction frame data; sequentially acquiring grids from the prediction frame data table;

Accordingly, the eleventh execution unit is configured to execute an eleventh operation on the mesh acquired by the tenth acquisition unit:

accordingly, an eleventh execution unit including a twelfth acquisition unit and a twelfth execution unit;

correspondingly, a twelfth acquisition unit is used for sequentially acquiring the predicted frame data from the second predicted frame data corresponding to the grid acquired by the eleventh acquisition unit, and triggering a twelfth execution unit;

accordingly, a twelfth execution unit for executing a twelfth operation on the prediction frame data acquired by the twelfth acquisition unit; a twelfth execution unit, specifically configured to obtain a first confidence parameter from the predicted frame data obtained by the twelfth obtaining unit, and mark the predicted frame data in the predicted frame data table when the first confidence parameter is fifth preset data;

accordingly, eleven output units for outputting all marked prediction frame data in the prediction frame data table.

Further, the twelfth execution unit further includes: a save acquisition subunit and a thirteenth execution operation unit,

Further correspondingly, a saving and acquiring subunit, configured to save the prediction frame data to a twelfth list; sequentially acquiring prediction frame data from a twelfth list to trigger a thirteenth execution unit;

Further, correspondingly, a thirteenth execution operation unit for executing a thirteenth operation on the prediction frame data;

further, a thirteenth execution operation unit is specifically configured to obtain all the category probability data from the prediction frame data of the storage obtaining subunit, select the largest category probability data from all the obtained category probability data, determine the category data of the prediction frame data according to the largest category probability data, and store the category data and the prediction frame data in correspondence in twelve lists; classifying all prediction frame data in the twelve lists according to all the category data in the twelve lists; any category is acquired from the twelve lists, a fourteenth execution operation unit is triggered, and when all the categories are acquired from the twelve lists, the twelfth list is emptied;

further, correspondingly, a fourteenth execution operation unit for executing a fourteenth operation on the category acquired by the thirteenth execution operation unit;

further correspondingly, a fourteenth execution operation unit is specifically configured to acquire prediction frame data corresponding to the category from the twelfth list, and store the prediction frame data in the thirteenth list; acquiring the largest category probability data in all category probability data corresponding to the category from the prediction frame data in the thirteenth list; marking prediction frame data corresponding to the maximum category probability data in a thirteenth list; according to an eleventh preset algorithm, calculating the predicted frame data corresponding to the maximum category probability data and other predicted frame data in the thirteenth list respectively to obtain sixth confidence parameters corresponding to the other predicted frame data in the thirteenth list respectively; performing a fifteenth operation on each sixth confidence parameter; the fifteenth operation is: the device compares the sixth confidence parameter with the size of the sixth comparison data, and marks the predicted frame data in the thirteenth list when the sixth confidence parameter is smaller than the thirteenth confidence data; correspondingly marking all marked predicted frame data in the thirteenth list in a predicted frame data table; clearing the thirteenth list;

Further, when the fourteenth execution operation unit is specifically configured to calculate, according to the eleventh preset algorithm, the predicted frame data corresponding to the maximum class probability data and the other predicted frame data in the thirteenth list to obtain sixth confidence parameters corresponding to the other predicted frame data in the thirteenth list, respectively, the fourteenth execution operation unit is specifically configured to obtain, from the thirteenth list, the predicted frame data corresponding to the maximum class probability data, and record the predicted frame data as labeled predicted frame data; when the predicted frame data except the marked predicted frame data is obtained from the thirteenth list, the device calculates the marked predicted frame data and the second current predicted frame data according to an eleventh preset algorithm to obtain an intersection area and a union area, and calculates the intersection area and the union area according to a first preset algorithm to obtain a sixth confidence parameter.

The foregoing is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions easily contemplated by those skilled in the art within the technical scope of the present invention should be included in the scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for image recognition based on a deep learning network is characterized in that;

2. The method of claim 1, wherein the grid data comprises a grid abscissa, a grid ordinate, a grid length, and a grid width;

the prediction frame data table comprises fourth data, wherein the prediction frame data comprises a prediction frame abscissa, a prediction frame ordinate, a prediction frame length, a prediction frame width, a first confidence parameter and third preset data category probability data;

and the fourth data has binding relation with the first preset data and the second preset data.

3. The method of claim 2, wherein the binding relationship is: the fourth data is the product of the first preset data and the second preset data;

the first preset data is in the format of data S.

4. The method of claim 1, wherein the third operation comprises the steps of:

step D01: means for obtaining category data of the image from actual coordinate data of the image;

5. The method of claim 1, wherein the fourth operation comprises the steps of:

step D11: the device calculates the intersection area and the union area of the predicted frame data and the image according to eleven preset algorithms on the predicted frame data and the actual data of the image; calculating the intersection area and the union area to obtain a second confidence parameter;

step D12: the device acquires a first confidence parameter in the predicted frame data, and determines the confidence coefficient of the predicted frame data according to the first confidence parameter, the second confidence parameter, the fifth preset data and the sixth preset data; and correspondingly storing the confidence degree of the predicted frame data, the predicted frame data and the category data of the predicted frame data into a first list.

6. The method according to claim 5, wherein the step D12 is specifically: the device acquires a first confidence parameter in the predicted frame data, determines a category of the first confidence parameter, and when the first confidence parameter is fifth preset data, determines the confidence coefficient of the predicted frame data according to the fifth preset data and the second confidence parameter, and correspondingly stores the confidence coefficient of the predicted frame data, the category data and the actual data confidence coefficient of the image into a first list; and when the first confidence parameter is sixth preset data, determining the confidence coefficient of the predicted frame data according to the second confidence parameter, and correspondingly storing the confidence coefficient of the predicted frame data, the category data and the actual data of the image into a first list.

7. The method according to claim 6, wherein the step D12 is specifically: the device acquires a first confidence parameter in the predicted frame data, determines the category of the first confidence parameter, and when the first confidence parameter is fifth preset data, calculates the fifth preset data and the second confidence parameter according to a second preset algorithm to obtain the confidence coefficient of the predicted frame data, and correspondingly stores the confidence coefficient of the predicted frame data, the category data and the actual data of the image into a first list to a first list; and when the first confidence parameter is sixth preset data, recording the second confidence parameter as the confidence coefficient of the predicted frame data, and correspondingly storing the confidence coefficient of the predicted frame data, the category data and the actual data of the image into a first list.

8. The method according to claim 5, wherein in the step D11, the device obtains an intersection area and a union area of the prediction frame data and the image according to the prediction frame data and the actual data of the image, specifically: and calculating the abscissa, the ordinate, the length and the width of the predicted frame data and the actual abscissa, the actual ordinate, the actual length and the actual width in the actual data of the image according to an eleven preset algorithm to obtain the intersection area and the union area of the predicted frame corresponding to the first current predicted frame data and the second current image.

9. The method of claim 5, wherein in the step D11, the calculating the intersection area and the union area obtains a second confidence parameter, specifically: the device calculates the intersection area and the union area according to a first preset algorithm to obtain a second confidence parameter.

10. The method of claim 9, wherein in the step D11, the calculating the intersection area and the union area obtains a second confidence parameter, specifically: and the device performs comparison operation on the intersection area and the union area according to a first preset algorithm to obtain a second confidence parameter.

11. The method according to claim 1, wherein in the step S1-03, the predicted frame data corresponding to the maximum confidence level and the actual coordinate data of the image obtained from the first list are correspondingly stored in the second list, specifically: acquiring prediction frame data corresponding to the maximum confidence coefficient and actual coordinate data of the image from the first list, and correspondingly storing the maximum confidence coefficient, the prediction frame data and the actual coordinate data of the image into the second list; the device acquires a fourth list from the second list;

step D22: the device obtains the maximum confidence from the third list, marks the predicted frame data corresponding to the maximum confidence, and marks the predicted frame data corresponding to the maximum confidence as first marked predicted frame data;

step D23: the device selects one prediction frame data except the first marked prediction frame data from the third list as second current prediction frame data;

step D24: the device acquires actual data corresponding to the second current prediction frame data from a third list; obtaining the intersection area and the union area of the prediction frames corresponding to the first marked prediction frame data and the prediction frames corresponding to the second current prediction frame data according to the prediction frame data corresponding to the first marked prediction frame data and the prediction frame data of the second current prediction frame; calculating the intersection area and the union area to obtain a third confidence parameter;

step D27: the device acquires next predicted frame data from the predicted frames except the first marked predicted frame data in the third list to serve as second current predicted frame data, and returns to the step D24;

Step D29: the device selects the next category from the categories in the second list as the first current category and returns to step D21.

12. The method of claim 1, wherein the fifth operation comprises the steps of: and the device calculates deviation data of the predicted frame data and the actual coordinate data according to the predicted frame data and the actual coordinate data, and stores the deviation data into a fifth list.

13. The method according to claim 1, wherein in the step S3, the processing the picture to be identified using the deep learning network containing the first parameter data to obtain the prediction frame data table includes the following steps:

step T1: when a picture to be identified is acquired, the device divides the picture into first preset data grids according to the first preset data; identifying the picture to be identified by using a deep learning network comprising updated first parameter data according to the first preset data, the second preset data and the third preset data to obtain a prediction frame data table; the prediction frame data table comprises second preset data which correspond to all grids respectively and prediction frame data; sequentially obtaining grids from the prediction frame data table, and executing eleventh operation on the grids:

step T2: the apparatus outputs all marked prediction frame data in the prediction frame data table.

14. The method of claim 13, wherein in the twelfth operation, marking the prediction block data in a prediction block data table is replaced with: means for saving the prediction frame data to a twelfth list; sequentially obtaining prediction frame data from a twelfth list, and executing thirteenth operation on the prediction frame data;

when one prediction frame data is obtained from the twelfth list, the device obtains all the category probability data from the prediction frame data, selects the largest category probability data from the obtained all the category probability data, determines the category data of the prediction frame data according to the largest category probability data, and correspondingly stores the category data and the prediction frame data in the twelfth list; classifying all prediction frame data in the twelve lists according to all the category data in the twelve lists; any category is acquired from the twelve lists, a fourteenth operation is executed, and when all the categories are acquired from the twelve lists, the twelfth list is emptied;

The fourteenth operation is: the device acquires the predicted frame data corresponding to the category from the twelfth list and stores the predicted frame data into the thirteenth list; acquiring the largest category probability data in all category probability data corresponding to the category from the prediction frame data in the thirteenth list; marking prediction frame data corresponding to the maximum category probability data in a thirteenth list; according to an eleventh preset algorithm, calculating the predicted frame data corresponding to the maximum category probability data and other predicted frame data in the thirteenth list respectively to obtain sixth confidence parameters corresponding to the other predicted frame data in the thirteenth list respectively; performing a fifteenth operation on each sixth confidence parameter; the fifteenth operation is: the apparatus compares the sixth confidence parameter with the size of the sixth comparison data, and marks the predicted box data in a thirteenth list when the sixth confidence parameter is less than the thirteenth confidence data; correspondingly marking all marked predicted frame data in the thirteenth list in a predicted frame data table; the thirteenth list is emptied.

15. The method of claim 14, wherein the calculating, according to the eleventh preset algorithm, the prediction frame data corresponding to the maximum class probability data and the other prediction frame data in the thirteenth list respectively obtain sixth confidence parameters corresponding to the other prediction frame data in the thirteenth list respectively, specifically includes:

16. The device for carrying out image recognition based on the deep learning network is characterized by comprising an incoming module, an acquisition module, an execution module, a deviation comparison module, a module to be recognized and a parameter updating module;

the execution module comprises a first execution operation unit;

the obtaining module is further configured to, when all the pictures in the picture sample library are obtained, obtain prediction frame data and actual coordinate data of the corresponding image from the second list, and trigger a fifth execution operation unit;

17. The apparatus of claim 16, wherein the grid data comprises a grid abscissa, a grid ordinate, a grid length, and a grid width;

18. The apparatus of claim 17, wherein the binding relationship is: the fourth data is the product of the first preset data and the second preset data;

The first preset data is the product of data S and data S.

19. The apparatus of claim 16, wherein the save acquisition unit comprises an acquisition as a subunit, an acquisition subunit, a first determination subunit, a save subunit, a second determination subunit;

the acquisition serves as a subunit, and is used for acquiring one image from all images corresponding to the grid to serve as an image;

the acquiring subunit is configured to acquire actual coordinate data of the image that is acquired as the subunit, and acquire category data of the image from the actual coordinate data of the image that is acquired as the subunit;

the first judging subunit is configured to judge whether the category data of the image obtained as the subunit is the same as the category data of the prediction frame data, and if yes, trigger the storing subunit;

the storage subunit is configured to store the actual coordinate data of the image and the prediction frame data that are obtained as the subunits into a prediction frame same-category image table, trigger the second determination subunit, and otherwise trigger the second determination subunit;

the second judging subunit is configured to judge whether a next image can be acquired from all the images corresponding to the grid as an image, if yes, acquire the next image from all the images corresponding to the grid as an image, and trigger the acquiring subunit; otherwise, acquiring images from the image table of the same category corresponding to the prediction frame in sequence, and triggering the fourth operation execution unit.

20. The apparatus of claim 16, wherein the fourth execution operation unit comprises a first calculation subunit and a first storage subunit;

the first calculating subunit is configured to obtain an intersection area and a union area of the prediction frame data and the image according to the eleven preset algorithm for the prediction frame data and the actual data of the image acquired by the second acquiring unit; calculating the intersection area and the union area to obtain a second confidence parameter;

the first storage subunit is configured to obtain a first confidence parameter in the prediction frame data, and determine a confidence coefficient of the prediction frame data according to the first confidence parameter, the second confidence parameter of the first calculation subunit, the fifth preset data and the sixth preset data; and correspondingly storing the confidence degree of the predicted frame data, the predicted frame data and the category data of the predicted frame data into a first list.

21. The apparatus of claim 20, wherein the first storage subunit is specifically configured to obtain a first confidence parameter in the prediction frame data, determine a class of the first confidence parameter, and when the first confidence parameter is fifth preset data, determine a confidence level of the prediction frame data according to the fifth preset data and the second confidence parameter, and correspondingly store the confidence level of the prediction frame data, the class data, and an actual data confidence level of the image to a first list; and when the first confidence parameter is sixth preset data, determining the confidence coefficient of the predicted frame data according to the second confidence parameter, and correspondingly storing the confidence coefficient of the predicted frame data, the category data and the actual data of the image into a first list.

22. The apparatus of claim 21, wherein the first storage subunit is specifically configured to obtain a first confidence parameter in the predicted frame data, determine a class of the first confidence parameter, and calculate, when the first confidence parameter is fifth preset data, the fifth preset data and the second confidence parameter according to a second preset algorithm to obtain a confidence level of the predicted frame data, and correspondingly store the confidence level of the predicted frame data, the class data, and the actual data of the image into a first list to a first list; and when the first confidence parameter is sixth preset data, recording the second confidence parameter as the confidence coefficient of the predicted frame data, and correspondingly storing the confidence coefficient of the predicted frame data, the category data and the actual data of the image into a first list.

23. The apparatus of claim 21, wherein when the first calculating subunit is configured to obtain an intersection area and a union area of the predicted frame data and the image according to the eleven preset algorithm on the predicted frame data and the actual data of the image acquired by the second acquiring unit, the first calculating subunit is configured to calculate an intersection area and a union area of a predicted frame corresponding to the first current predicted frame data and the second current image according to the eleven preset algorithm on a predicted frame abscissa, a predicted frame ordinate, a predicted frame length, and a predicted frame width of the predicted frame data and an actual abscissa, an actual ordinate, an actual length, and an actual width of the actual data of the image.

24. The apparatus of claim 21, wherein when the first computing subunit is configured to operate on the intersection area and the union area to obtain the second confidence parameter, the first computing subunit is specifically configured to operate on the intersection area and the union area according to a first preset algorithm to obtain the second confidence parameter.

25. The apparatus of claim 24, wherein when the first computing subunit is configured to perform a comparison of the intersection area and the union area to obtain the second confidence parameter, the first computing subunit is configured to perform a comparison of the intersection area and the union area to obtain the second confidence parameter.

26. The apparatus of claim 16, wherein when the judgment saving and emptying unit is configured to correspondingly save the predicted frame data and the actual coordinate data of the image corresponding to the maximum confidence level from the first list to the second list, the judgment saving and emptying unit is specifically configured to correspondingly save the predicted frame data and the actual coordinate data of the image corresponding to the maximum confidence level from the first list to the second list;

The acquisition module further comprises a first acquisition unit, wherein the first acquisition unit is used for acquiring a fourth list from the second list;

the first acquisition unit comprises a first classification subunit, a first storage subunit, a first marking subunit, a comparison subunit, a first acquisition subunit and a second storage subunit;

the first classifying subunit is configured to classify the prediction frame data in the second list, and sequentially select one category from the obtained categories as a first current category;

the first storing subunit is configured to obtain prediction frame data corresponding to the first current category obtained by the first classification subunit from a second list, and store actual data of an image and confidence level of the prediction frame data in a third list;

the first marking subunit is configured to obtain a maximum confidence coefficient from the third list, mark prediction frame data corresponding to the maximum confidence coefficient, and mark the prediction frame data corresponding to the maximum confidence coefficient as first marked prediction frame data;

the first obtaining subunit is further configured to select, from the third list, one prediction frame data other than the first marked prediction frame data as second current prediction frame data;

The second storage subunit is configured to obtain actual data corresponding to the second current prediction frame data from the third list; obtaining the intersection area and the union area of the prediction frames corresponding to the first marked prediction frame data and the prediction frames corresponding to the second current prediction frame data according to the prediction frame data corresponding to the first marked prediction frame data and the prediction frame data of the second current prediction frame; calculating the intersection area and the union area to obtain a third confidence parameter;

the comparison subunit is used for comparing the third confidence parameter obtained by the confidence parameter storage subunit with the third comparison data; marking a second current prediction frame in a third list when the third confidence parameter is less than the third comparison data;

the first storage subunit is further configured to, when the first obtaining unit obtains that all the prediction frame data is selected from the third list except the first marked prediction frame data, store the marked prediction frame in the third list and the corresponding confidence, actual data, prediction frame data, and category data in the fourth list, and empty the third list.

27. The apparatus of claim 16, wherein the fifth execution operation unit is configured to calculate deviation data of the predicted frame data and the actual coordinate data from the predicted frame data and the actual coordinate data, and store the deviation data to a fifth list.

28. The apparatus of claim 16, wherein the processing module to be identified comprises an eleventh acquisition unit, an eleventh execution unit, and an eleventh output unit;

the eleventh acquisition unit is used for dividing the picture into first preset data grids according to the first preset data; identifying the picture to be identified by using a deep learning network comprising updated first parameter data according to the first preset data, the second preset data and the third preset data to obtain a prediction frame data table; the prediction frame data table comprises second preset data which correspond to all grids respectively and prediction frame data; sequentially acquiring grids from the prediction frame data table;

the eleventh execution unit is configured to execute an eleventh operation on the grid acquired by the eleventh acquisition unit:

the eleventh execution unit comprises a twelfth acquisition unit and a twelfth execution unit;

the twelfth acquisition unit is used for sequentially acquiring prediction frame data from second preset data corresponding to the grid acquired by the eleventh acquisition unit and triggering the twelfth execution unit;

the twelfth execution unit is configured to execute a twelfth operation on the prediction frame data acquired by the twelfth acquisition unit; the twelfth execution unit is specifically configured to obtain a first confidence parameter from the predicted frame data obtained by the twelfth obtaining unit, and mark the predicted frame data in a predicted frame data table when the first confidence parameter is fifth preset data;

And the eleven output unit is used for outputting all marked predicted frame data in the predicted frame data table.

29. The apparatus of claim 28, wherein the twelfth execution unit further comprises: a save acquisition subunit, a thirteenth execution operation unit, and a fourteenth execution operation unit;

the storage and acquisition subunit is configured to store the prediction frame data to a twelfth list; sequentially acquiring prediction frame data from a twelfth list to trigger the thirteenth execution operation unit;

the thirteenth execution operation unit is configured to execute a thirteenth operation on the prediction frame data;

the thirteenth execution operation unit is specifically configured to obtain all the category probability data from the prediction frame data in the save and obtain subunit, select the largest category probability data from all the obtained category probability data, determine the category data of the prediction frame data according to the largest category probability data, and save the category data and the prediction frame data in a twelve-list manner; classifying all prediction frame data in the twelve lists according to all the category data in the twelve lists; any category is acquired from the twelve lists, the fourteenth execution operation unit is triggered, and when all categories are acquired from the twelve lists, the twelfth list is emptied;

The thirteenth execution operation unit is configured to execute a thirteenth operation on the category acquired by the thirteenth execution operation unit;

the fourteenth execution operation unit is specifically configured to obtain, from the twelfth list, prediction frame data corresponding to the category, and store the prediction frame data in the thirteenth list; acquiring the largest category probability data in all category probability data corresponding to the category from the prediction frame data in the thirteenth list; marking prediction frame data corresponding to the maximum category probability data in a thirteenth list; according to an eleventh preset algorithm, calculating the predicted frame data corresponding to the maximum category probability data and other predicted frame data in the thirteenth list respectively to obtain sixth confidence parameters corresponding to the other predicted frame data in the thirteenth list respectively; performing a fifteenth operation on each sixth confidence parameter; the fifteenth operation is: the apparatus compares the sixth confidence parameter with the size of the sixth comparison data, and marks the predicted box data in a thirteenth list when the sixth confidence parameter is less than the thirteenth confidence data; correspondingly marking all marked predicted frame data in the thirteenth list in a predicted frame data table; the thirteenth list is emptied.

30. The apparatus of claim 29, wherein when the fourteenth execution operation unit is specifically configured to calculate, according to an eleventh preset algorithm, prediction frame data corresponding to the maximum class probability data and other prediction frame data in the thirteenth list to obtain sixth confidence parameters corresponding to the other prediction frame data in the thirteenth list, respectively, the fourteenth execution operation unit is specifically configured to obtain, from the thirteenth list, the prediction frame data corresponding to the maximum class probability data, and record the prediction frame data as labeled prediction frame data; when the predicted frame data except the marked predicted frame data is obtained from the thirteenth list, the device calculates the marked predicted frame data and the second current predicted frame data according to an eleventh preset algorithm to obtain an intersection area and a union area, and calculates the intersection area and the union area according to a first preset algorithm to obtain a sixth confidence parameter.