CN110443118B - Commodity identification method, system and medium based on artificial features - Google Patents
Commodity identification method, system and medium based on artificial features Download PDFInfo
- Publication number
- CN110443118B CN110443118B CN201910549890.4A CN201910549890A CN110443118B CN 110443118 B CN110443118 B CN 110443118B CN 201910549890 A CN201910549890 A CN 201910549890A CN 110443118 B CN110443118 B CN 110443118B
- Authority
- CN
- China
- Prior art keywords
- commodity
- artificial
- result
- decoding
- positioning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
- G06K7/00—Methods or arrangements for sensing record carriers, e.g. for reading patterns
- G06K7/10—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation
- G06K7/10544—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation by scanning of the records by radiation in the optical part of the electromagnetic spectrum
- G06K7/10712—Fixed beam scanning
- G06K7/10722—Photodetector array or CCD scanning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/35—Categorising the entire scene, e.g. birthday party or wedding scene
- G06V20/36—Indoor scenes
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07G—REGISTERING THE RECEIPT OF CASH, VALUABLES, OR TOKENS
- G07G1/00—Cash registers
- G07G1/0036—Checkout procedures
- G07G1/0045—Checkout procedures with a code reader for reading of an identifying code of the article to be registered, e.g. barcode reader or radio-frequency identity [RFID] reader
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07G—REGISTERING THE RECEIPT OF CASH, VALUABLES, OR TOKENS
- G07G1/00—Cash registers
- G07G1/12—Cash registers electronically operated
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Electromagnetism (AREA)
- Toxicology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a commodity identification method, a system and a medium based on artificial features, which comprises the following steps: an image acquisition step: shooting commodities in a sales counter, triggering image acquisition when a door of the sales counter is opened and closed, and uploading images acquired by the image acquisition; a commodity positioning step: according to the uploaded picture, positioning the feature to be recognized by using the obtained feature training model to obtain a positioning result; and (3) commodity identification and clearing step: and identifying and clearing according to the obtained positioning result and the characteristic to be identified. The invention directly positions and identifies the commodities with obvious characteristics through the neural network, and solves the problems of too complex algorithm and too long algorithm time.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a commodity recognition method, a commodity recognition system and a commodity recognition medium based on artificial features.
Background
A paper "two-dimensional code region extraction [ J ] based on BP neural network. microcomputer and application 2015,34(01):50-52+ 58" proposes a two-dimensional code image positioning, identifying and preprocessing system, which can quickly find a two-dimensional code region. The system is mainly divided into four parts: image processing, feature extraction, sample training and sample testing. The image processing includes image enhancement and image opening and closing operation operations. The working principle of the system is as follows: the original image is changed into a binary image through image enhancement and opening and closing operations, the binary image comprises one or more rectangular areas, and the interference areas are filtered through a BP neural network to obtain a two-dimensional code area.
In the paper, the BP neural network is used for filtering the interference area instead of directly positioning the two-dimensional code, and the accuracy of filtering cannot be guaranteed. In practical situations, even if the system locates the two-dimensional code accurately, the system cannot decode the two-dimensional code correctly due to lack of post-processing.
CN109145816A (application number: 201810953349.5) discloses a commodity identification method and a system, relating to the field of commodity identification. The commodity identification method comprises the following steps: the neural network module acquires commodity characteristics and transmits the commodity characteristics to the channel domain attention module, wherein the commodity characteristics comprise commodity-related characteristics and commodity-unrelated characteristics; and the channel domain attention module distinguishes the commodity-related feature and the commodity-unrelated feature and at least transmits the commodity-related feature to the next neural network module.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a commodity identification method, a commodity identification system and a commodity identification medium based on artificial characteristics.
The invention provides a commodity identification method based on artificial features, which comprises the following steps:
an image acquisition step: shooting commodities in a sales counter, triggering image acquisition when a door of the sales counter is opened and closed, and uploading images acquired by the image acquisition;
a commodity positioning step: according to the uploaded picture, positioning the feature to be recognized by using the obtained feature training model to obtain a positioning result;
and (3) commodity identification and clearing step: and identifying and clearing according to the obtained positioning result and the characteristic to be identified.
Preferably, the method further comprises the following steps:
a characteristic preparation step: making artificial characteristics to be attached to commodities and putting the commodities in a sales counter;
model training: carrying out neural network training on the commodity features of the commodities with obvious features to obtain a commodity training model with obvious features; and for the commodities with the artificial characteristics, carrying out neural network training on the artificial characteristics of the commodities to obtain artificial characteristic commodity training models.
Preferably, the feature training model comprises: the system comprises an obvious characteristic commodity training model and an artificial characteristic commodity training model;
the features to be identified include: commodity features, artificial features.
Preferably, the goods identification clearing step includes:
and a positioning result judging step: judging whether the positioning result is empty: if the positioning result is null, the step of checking the whole picture is carried out continuously, and if the positioning result is not null, the step of identifying the commodity is carried out continuously
A commodity identification step: according to the obtained positioning result, identifying the commodity characteristics and the artificial characteristics to obtain a commodity identification result and an artificial characteristic identification result, if the artificial characteristic identification result is empty, obtaining the commodity identification result and entering the step of overall image review to continue execution, and if the artificial characteristic identification result is not empty, obtaining the commodity identification result and entering the step of commodity counting to continue execution;
and (3) a whole picture review step: performing distortion processing on the whole image, then decoding according to an industrial decoding standard, taking a decoding result as a final commodity identification result, and entering a commodity counting step;
and (4) commodity counting: and comparing the commodity information in the sales counter at the two times of opening and closing the door according to the obtained commodity identification result and the artificial characteristic identification result to obtain a final result of commodity identification.
Preferably, the feature preparation step:
the artificial features include: two-dimensional codes and dot matrix codes;
the sales counter comprises: a sales counter (1), a camera (2) and an illuminating lamp (3);
the camera (2), the illuminating lamp (3) and the commodity (4) are all arranged in the sales counter (1), and the camera is arranged above the commodity (4) in the sales counter (1) and shoots the characteristics of the commodity and artificial characteristics on part of the commodity.
Preferably, the article positioning step:
positioning by using an artificial characteristic commodity training model according to the uploaded picture to obtain coordinates of all artificial characteristics in the picture, framing the artificial characteristics according to the coordinates of all artificial characteristics, intercepting the picture of the content in the frame according to the positioning frame, and outputting a positioning result and a screenshot;
and for the commodities without the adhered artificial characteristics, positioning the commodities by using the commodity training model with obvious characteristics, and outputting a positioning result.
And if the positioning result is null, the step of reviewing the whole image is continued, and if the positioning result is not null, the step of identifying the commodity is continued.
Preferably, the article identifying step:
for commodities without artificial features, identifying by using a commodity training model with obvious features according to the obtained positioning result to obtain a commodity identification result;
for the commodity pasted with the artificial features, decoding the obtained screenshot, wherein the decoding comprises the following steps:
performing opening operation and closing operation on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a first decoding result;
carrying out contrast enhancement on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a second decoding result;
carrying out distortion correction on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a second decoding result;
according to the obtained first, second and third decoding results, if two or more same results appear in the three decoding results of each screenshot, the decoding result is regarded as an artificial feature recognition result; if only one decoding result appears and the other two methods cannot decode, taking the decoding result as an artificial feature identification result;
judging whether the artificial characteristic identification result is empty: if the artificial characteristic identification result is empty, entering a full-image review step to continue execution; if the artificial characteristic identification result is not null, obtaining a commodity identification result and entering a commodity counting step for continuous execution;
judging whether the artificial feature recognition result is empty:
and if the following conditions occur, judging that the artificial feature recognition result is empty: all three methods cannot be decoded; one decoding result is null, and the other two decoding results are different; the three decoding results are all different;
the opening operation and the closing operation:
the opening and closing operation is a morphological processing method, and the operation of firstly corroding and then expanding is called as opening operation;
the closed operation refers to an operation of expansion followed by corrosion.
According to the invention, the commodity identification system based on artificial features comprises:
an image acquisition module: shooting commodities in a sales counter, triggering image acquisition when a door of the sales counter is opened and closed, and uploading images acquired by the image acquisition;
a commodity positioning module: according to the uploaded picture, positioning the feature to be recognized by using the obtained feature training model to obtain a positioning result;
the commodity identification and clearing module: and identifying and clearing according to the obtained positioning result and the characteristic to be identified.
Preferably, the method further comprises the following steps:
a feature preparation module: making artificial characteristics to be attached to commodities and putting the commodities in a sales counter;
a model training module: carrying out neural network training on the commodity features of the commodities with obvious features to obtain a commodity training model with obvious features; for the commodities with the artificial characteristics, carrying out neural network training on the artificial characteristics of the commodities to obtain artificial characteristic commodity training models;
the feature training model comprises: the system comprises an obvious characteristic commodity training model and an artificial characteristic commodity training model;
the features to be identified include: commodity characteristics, artificial characteristics;
the commodity identification and clearing module comprises:
a positioning result judging module: judging whether the positioning result is empty: if the positioning result is null, the system enters a full-image review module to continue execution, and if the positioning result is not null, the system enters a commodity identification module to continue execution
A commodity identification module: according to the obtained positioning result, identifying the commodity characteristics and the artificial characteristics to obtain a commodity identification result and an artificial characteristic identification result, if the artificial characteristic identification result is empty, obtaining the commodity identification result and calling a full-image review module, and if the artificial characteristic identification result is not empty, obtaining the commodity identification result and calling a commodity counting module;
the whole-image review module: performing distortion processing on the whole image, then decoding according to an industrial decoding standard, taking a decoding result as a final commodity identification result, and calling a commodity counting module;
the commodity counting module: according to the obtained commodity identification result and the artificial characteristic identification result, comparing the commodity information in the sales counter at the two moments of opening and closing the door to obtain the final result of commodity identification;
the feature preparation module:
the artificial features include: two-dimensional codes and dot matrix codes;
the sales counter comprises: a sales counter (1), a camera (2) and an illuminating lamp (3);
the camera (2), the illuminating lamp (3) and the commodity (4) are all arranged in the sales counter (1), and the camera is arranged above the commodity (4) in the sales counter (1) and is used for shooting the characteristics of the commodity and artificial characteristics on part of the commodity;
the commodity positioning module:
positioning by using an artificial characteristic commodity training model according to the uploaded picture to obtain coordinates of all artificial characteristics in the picture, framing the artificial characteristics according to the coordinates of all artificial characteristics, intercepting the picture of the content in the frame according to the positioning frame, and outputting a positioning result and a screenshot;
and for the commodities without the adhered artificial characteristics, positioning the commodities by using the commodity training model with obvious characteristics, and outputting a positioning result.
If the positioning result is null, calling a full-image review module, and if the positioning result is not null, calling a commodity identification module;
the commodity identification module:
for commodities without artificial features, identifying by using a commodity training model with obvious features according to the obtained positioning result to obtain a commodity identification result;
for the commodity pasted with the artificial features, decoding the obtained screenshot, wherein the decoding comprises the following steps:
performing opening operation and closing operation on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a first decoding result;
carrying out contrast enhancement on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a second decoding result;
carrying out distortion correction on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a second decoding result;
according to the obtained first, second and third decoding results, if two or more same results appear in the three decoding results of each screenshot, the decoding result is regarded as an artificial feature recognition result; if only one decoding result appears and the other two methods cannot decode, taking the decoding result as an artificial feature identification result;
judging whether the artificial characteristic identification result is empty: if the artificial characteristic identification result is null, calling a full-image review module; if the artificial characteristic identification result is not null, obtaining a commodity identification result and calling a commodity counting module;
judging whether the artificial feature recognition result is empty:
and if the following conditions occur, judging that the artificial feature recognition result is empty: all three methods cannot be decoded; one decoding result is null, and the other two decoding results are different; the three decoding results are all different;
the opening operation and the closing operation:
the opening and closing operation is a morphological processing method, and the operation of firstly corroding and then expanding is called as opening operation;
the closed operation refers to an operation of expansion followed by corrosion.
According to the present invention, there is provided a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of any of the above-described artificial characteristic-based article identification methods.
Compared with the prior art, the invention has the following beneficial effects:
1. the invention directly positions and identifies the commodities with obvious characteristics through the neural network, and solves the problems of too complex algorithm and too long algorithm time.
2. The invention adds artificial characteristics to commodities with unobvious characteristics, and identifies the commodities through the medium of the artificial characteristics, thereby solving the problem that part of the commodities can not be positioned and identified.
3. By adopting various methods such as machine vision, image processing and the like, compared with the method of directly identifying according to the industrial standard, the method solves the problems of low understanding code rate and low decoding accuracy.
Drawings
Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of non-limiting embodiments with reference to the following drawings:
FIG. 1 is a schematic structural diagram of a sales counter according to the present invention.
Fig. 2 is a schematic diagram of a work flow of the system provided by the present invention.
Fig. 3 is a schematic diagram of the algorithm flow provided by the present invention.
Detailed Description
The present invention will be described in detail with reference to specific examples. The following examples will assist those skilled in the art in further understanding the invention, but are not intended to limit the invention in any way. It should be noted that it would be obvious to those skilled in the art that various changes and modifications can be made without departing from the spirit of the invention. All falling within the scope of the present invention.
The invention provides a commodity identification method based on artificial features, which comprises the following steps:
an image acquisition step: shooting commodities in a sales counter, triggering image acquisition when a door of the sales counter is opened and closed, and uploading images acquired by the image acquisition;
a commodity positioning step: according to the uploaded picture, positioning the feature to be recognized by using the obtained feature training model to obtain a positioning result;
and (3) commodity identification and clearing step: and identifying and clearing according to the obtained positioning result and the characteristic to be identified.
Specifically, the method further comprises the following steps:
a characteristic preparation step: making artificial characteristics to be attached to commodities and putting the commodities in a sales counter;
model training: carrying out neural network training on the commodity features of the commodities with obvious features to obtain a commodity training model with obvious features; and for the commodities with the artificial characteristics, carrying out neural network training on the artificial characteristics of the commodities to obtain artificial characteristic commodity training models.
Specifically, the feature training model includes: the system comprises an obvious characteristic commodity training model and an artificial characteristic commodity training model;
the features to be identified include: commodity features, artificial features.
Specifically, the commodity identification clearing step includes:
and a positioning result judging step: judging whether the positioning result is empty: if the positioning result is null, the step of checking the whole picture is carried out continuously, and if the positioning result is not null, the step of identifying the commodity is carried out continuously
A commodity identification step: according to the obtained positioning result, identifying the commodity characteristics and the artificial characteristics to obtain a commodity identification result and an artificial characteristic identification result, if the artificial characteristic identification result is empty, obtaining the commodity identification result and entering the step of overall image review to continue execution, and if the artificial characteristic identification result is not empty, obtaining the commodity identification result and entering the step of commodity counting to continue execution;
and (3) a whole picture review step: performing distortion processing on the whole image, then decoding according to an industrial decoding standard, taking a decoding result as a final commodity identification result, and entering a commodity counting step;
and (4) commodity counting: and comparing the commodity information in the sales counter at the two times of opening and closing the door according to the obtained commodity identification result and the artificial characteristic identification result to obtain a final result of commodity identification.
Specifically, the feature preparation step:
the artificial features include: two-dimensional codes and dot matrix codes;
the sales counter comprises: a sales counter (1), a camera (2) and an illuminating lamp (3);
the camera (2), the illuminating lamp (3) and the commodity (4) are all arranged in the sales counter (1), and the camera is arranged above the commodity (4) in the sales counter (1) and shoots the characteristics of the commodity and artificial characteristics on part of the commodity.
Specifically, the commodity positioning step:
positioning by using an artificial characteristic commodity training model according to the uploaded picture to obtain coordinates of all artificial characteristics in the picture, framing the artificial characteristics according to the coordinates of all artificial characteristics, intercepting the picture of the content in the frame according to the positioning frame, and outputting a positioning result and a screenshot;
and for the commodities without the adhered artificial characteristics, positioning the commodities by using the commodity training model with obvious characteristics, and outputting a positioning result.
And if the positioning result is null, the step of reviewing the whole image is continued, and if the positioning result is not null, the step of identifying the commodity is continued.
Specifically, the commodity identification step:
for commodities without artificial features, identifying by using a commodity training model with obvious features according to the obtained positioning result to obtain a commodity identification result;
for the commodity pasted with the artificial features, decoding the obtained screenshot, wherein the decoding comprises the following steps:
performing opening operation and closing operation on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a first decoding result;
carrying out contrast enhancement on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a second decoding result;
carrying out distortion correction on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a second decoding result;
according to the obtained first, second and third decoding results, if two or more same results appear in the three decoding results of each screenshot, the decoding result is regarded as an artificial feature recognition result; if only one decoding result appears and the other two methods cannot decode, taking the decoding result as an artificial feature identification result;
judging whether the artificial characteristic identification result is empty: if the artificial characteristic identification result is empty, entering a full-image review step to continue execution; if the artificial characteristic identification result is not null, obtaining a commodity identification result and entering a commodity counting step for continuous execution;
judging whether the artificial feature recognition result is empty:
and if the following conditions occur, judging that the artificial feature recognition result is empty: all three methods cannot be decoded; one decoding result is null, and the other two decoding results are different; the three decoding results are all different;
the opening operation and the closing operation:
the opening and closing operation is a morphological processing method, and the operation of firstly corroding and then expanding is called as opening operation;
the closed operation refers to an operation of expansion followed by corrosion.
According to the invention, the commodity identification system based on artificial features comprises:
an image acquisition module: shooting commodities in a sales counter, triggering image acquisition when a door of the sales counter is opened and closed, and uploading images acquired by the image acquisition;
a commodity positioning module: according to the uploaded picture, positioning the feature to be recognized by using the obtained feature training model to obtain a positioning result;
the commodity identification and clearing module: and identifying and clearing according to the obtained positioning result and the characteristic to be identified.
Specifically, the method further comprises the following steps:
a feature preparation module: making artificial characteristics to be attached to commodities and putting the commodities in a sales counter;
a model training module: carrying out neural network training on the commodity features of the commodities with obvious features to obtain a commodity training model with obvious features; for the commodities with the artificial characteristics, carrying out neural network training on the artificial characteristics of the commodities to obtain artificial characteristic commodity training models;
the feature training model comprises: the system comprises an obvious characteristic commodity training model and an artificial characteristic commodity training model;
the features to be identified include: commodity characteristics, artificial characteristics;
the commodity identification and clearing module comprises:
a positioning result judging module: judging whether the positioning result is empty: if the positioning result is null, the system enters a full-image review module to continue execution, and if the positioning result is not null, the system enters a commodity identification module to continue execution
A commodity identification module: according to the obtained positioning result, identifying the commodity characteristics and the artificial characteristics to obtain a commodity identification result and an artificial characteristic identification result, if the artificial characteristic identification result is empty, obtaining the commodity identification result and calling a full-image review module, and if the artificial characteristic identification result is not empty, obtaining the commodity identification result and calling a commodity counting module;
the whole-image review module: performing distortion processing on the whole image, then decoding according to an industrial decoding standard, taking a decoding result as a final commodity identification result, and calling a commodity counting module;
the commodity counting module: according to the obtained commodity identification result and the artificial characteristic identification result, comparing the commodity information in the sales counter at the two moments of opening and closing the door to obtain the final result of commodity identification;
the feature preparation module:
the artificial features include: two-dimensional codes and dot matrix codes;
the sales counter comprises: a sales counter (1), a camera (2) and an illuminating lamp (3);
the camera (2), the illuminating lamp (3) and the commodity (4) are all arranged in the sales counter (1), and the camera is arranged above the commodity (4) in the sales counter (1) and is used for shooting the characteristics of the commodity and artificial characteristics on part of the commodity;
the commodity positioning module:
positioning by using an artificial characteristic commodity training model according to the uploaded picture to obtain coordinates of all artificial characteristics in the picture, framing the artificial characteristics according to the coordinates of all artificial characteristics, intercepting the picture of the content in the frame according to the positioning frame, and outputting a positioning result and a screenshot;
and for the commodities without the adhered artificial characteristics, positioning the commodities by using the commodity training model with obvious characteristics, and outputting a positioning result.
If the positioning result is null, calling a full-image review module, and if the positioning result is not null, calling a commodity identification module;
the commodity identification module:
for commodities without artificial features, identifying by using a commodity training model with obvious features according to the obtained positioning result to obtain a commodity identification result;
for the commodity pasted with the artificial features, decoding the obtained screenshot, wherein the decoding comprises the following steps:
performing opening operation and closing operation on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a first decoding result;
carrying out contrast enhancement on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a second decoding result;
carrying out distortion correction on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a second decoding result;
according to the obtained first, second and third decoding results, if two or more same results appear in the three decoding results of each screenshot, the decoding result is regarded as an artificial feature recognition result; if only one decoding result appears and the other two methods cannot decode, taking the decoding result as an artificial feature identification result;
judging whether the artificial characteristic identification result is empty: if the artificial characteristic identification result is null, calling a full-image review module; if the artificial characteristic identification result is not null, obtaining a commodity identification result and calling a commodity counting module;
judging whether the artificial feature recognition result is empty:
and if the following conditions occur, judging that the artificial feature recognition result is empty: all three methods cannot be decoded; one decoding result is null, and the other two decoding results are different; the three decoding results are all different;
the opening operation and the closing operation:
the opening and closing operation is a morphological processing method, and the operation of firstly corroding and then expanding is called as opening operation;
the closed operation refers to an operation of expansion followed by corrosion.
According to the present invention, there is provided a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of any of the above-described artificial characteristic-based article identification methods.
The present invention will be described more specifically below with reference to preferred examples.
Preferred example 1:
as shown in fig. 2, a commodity identification method based on artificial features includes:
step 1: the step 1 comprises the following steps:
step 1.1: in the preparation stage, artificial features (such as micro code two-dimensional codes) are manufactured and attached to the commodities, and the commodities are put into a sales counter with a specific structure, wherein the structure is shown in figure 1: comprises a sales counter 1, a camera 2, an illuminating lamp 3 and a commodity 4. The camera 2, the illuminating lamp 3 and the commodity 4 are all arranged in the sales counter 1. The camera is above the commodity 4 and shoots the characteristics of the goods and artificial characteristics on part of the commodities. The focal length, wide angle and the distance to goods parameter of camera 2 should combine specific conditions to set for, guarantee can clearly shoot the characteristics above the commodity.
Step 1.2: and (4) carrying out neural network training on the commodities with obvious characteristics to obtain a training model of the commodities, wherein the commodities do not have artificial characteristics.
Step 1.3: for the commodities added with the artificial characteristics, the artificial characteristics of the commodities are subjected to neural network training to obtain training models of the commodities.
Step 2: and (5) image acquisition. Camera 2 shoots commodity 4, opens and closes when the cabinet door of selling goods and all can trigger image acquisition, passes to cloud identification center on the picture and discerns.
And step 3: and (5) positioning. The step 3 comprises the following steps:
step 3.1: the cloud identification center obtains a picture to be identified.
Step 3.2: and (4) positioning by using the training model of the artificial features in the step 1.3 to obtain the coordinates of all the artificial features in the picture, namely preloading the training model, and loading the picture into a neural network. The network obtains the area with the picture prediction high probability as the feature code through processing, and outputs the coordinate value of the area. For the commodity without the pasted artificial features, the training model in the step 1.2 is used for directly positioning the commodity.
Step 3.3: and framing the artificial features according to the coordinates, and capturing the picture of the content in the frame according to the positioning frame.
Step 3.4: and if the neural network positioning results in the steps 3.1 and 3.2 are empty, directly entering the step 5.
And 4, step 4: and (5) an identification phase. The step 4 comprises the following steps:
step 4.1: and for the commodity without the artificial features, directly using the model in the step 1.2 and identifying according to the position in the step 3.2, and obtaining an identification result.
Step 4.2: for the commodity pasted with the artificial features, the following operations are performed on the screenshot in the step 3.3: (1) and performing opening operation and closing operation on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features. (2) And carrying out contrast enhancement on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features. (3) And carrying out distortion correction on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features. The opening and closing operation is a morphological processing method, and the operation of first etching and then expanding is called opening operation. It has the functions of eliminating fine objects, separating objects at fine parts and smoothing the boundaries of larger objects. The expansion followed by erosion operation is referred to as a close operation. It has the function of filling fine cavities in the object, connecting adjacent objects and smoothing the boundary.
Step 4.3: comparing the three decoding results in the step 4.2, and if two or more same results appear in the three results of each screenshot, considering the result as the decoding result of the artificial characteristic. If only one result appears and the other two methods cannot decode, the result is taken as the decoding result of the artificial characteristic. If the following conditions occur, the output is empty, and the step 5 is carried out, otherwise, the step 6 is carried out: (1) all three methods cannot be decoded. (2) One decoding result is null and the other two decoding results are different. (3) All three decoding results are different.
And 5: and (5) reviewing the whole picture. And (4) performing distortion conversion treatment on the whole image, and then directly decoding according to an industrial decoding standard, wherein the result is used as a final result after step 4.3 or step 3.4. The industrial decoding standard, for example, the artificial characteristic can be qr code two-dimensional code, the two-dimensional code has a decoding standard, and micro qr code two-dimensional code can also be pasted and also has a corresponding decoding standard. What artificial features are pasted, what decoding standard is used.
Step 6: and counting the commodities. And summarizing the identification results of all the commodities, returning the identification results to the service background, and then carrying out cloud processing. And comparing the information of the commodities in the cabinet at the two moments of opening and closing the door to obtain the final result of commodity identification.
Fig. 3 is a schematic diagram showing the algorithm flow from step 3 to step 6.
In the description of the present application, it is to be understood that the terms "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", and the like indicate orientations or positional relationships based on those shown in the drawings, and are only for convenience in describing the present application and simplifying the description, but do not indicate or imply that the referred device or element must have a specific orientation, be constructed in a specific orientation, and be operated, and thus, should not be construed as limiting the present application.
Those skilled in the art will appreciate that, in addition to implementing the systems, apparatus, and various modules thereof provided by the present invention in purely computer readable program code, the same procedures can be implemented entirely by logically programming method steps such that the systems, apparatus, and various modules thereof are provided in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Therefore, the system, the device and the modules thereof provided by the present invention can be considered as a hardware component, and the modules included in the system, the device and the modules thereof for implementing various programs can also be considered as structures in the hardware component; modules for performing various functions may also be considered to be both software programs for performing the methods and structures within hardware components.
The foregoing description of specific embodiments of the present invention has been presented. It is to be understood that the present invention is not limited to the specific embodiments described above, and that various changes or modifications may be made by one skilled in the art within the scope of the appended claims without departing from the spirit of the invention. The embodiments and features of the embodiments of the present application may be combined with each other arbitrarily without conflict.
Claims (6)
1. A commodity identification method based on artificial features is characterized by comprising the following steps:
an image acquisition step: shooting commodities in a sales counter, triggering image acquisition when a door of the sales counter is opened and closed, and uploading images acquired by the image acquisition;
a commodity positioning step: according to the uploaded picture, positioning the feature to be recognized by using the obtained feature training model to obtain a positioning result;
and (3) commodity identification and clearing step: identifying and clearing according to the obtained positioning result and the characteristic to be identified;
further comprising:
a characteristic preparation step: making artificial characteristics to be attached to commodities with unobvious characteristics, and putting the commodities into a sales counter;
model training: carrying out neural network training on the commodity features of the commodities with obvious features to obtain a commodity training model with obvious features; for the commodities with the artificial characteristics, carrying out neural network training on the artificial characteristics of the commodities to obtain artificial characteristic commodity training models;
the feature training model comprises: the system comprises an obvious characteristic commodity training model and an artificial characteristic commodity training model;
the features to be identified include: commodity characteristics, artificial characteristics;
the commodity identification and clearing step comprises the following steps:
and a positioning result judging step: judging whether the positioning result is empty: if the positioning result is empty, the step of reviewing the whole image is continued, and if the positioning result is not empty, the step of identifying the commodity is continued;
a commodity identification step: according to the obtained positioning result, identifying the commodity characteristics and the artificial characteristics to obtain a commodity identification result and an artificial characteristic identification result, if the artificial characteristic identification result is empty, obtaining the commodity identification result and entering the step of overall image review to continue execution, and if the artificial characteristic identification result is not empty, obtaining the commodity identification result and entering the step of commodity counting to continue execution; in the step of counting commodities, according to a commodity identification result and an artificial characteristic identification result, comparing commodity information in a sales counter at two moments of opening and closing a door to obtain a final result of commodity identification;
and (3) a whole picture review step: performing distortion processing on the whole image, then decoding according to an industrial decoding standard, taking a decoding result as a final commodity identification result, and entering a commodity counting step; in the step of counting the commodities, according to the obtained commodity identification result, comparing the commodity information in the sales counter at two moments of opening and closing the door, and obtaining the final result of the commodity identification.
2. The artificial characteristic-based commodity identification method according to claim 1, wherein the characteristic preparation step:
the artificial features include: two-dimensional codes and dot matrix codes;
the sales counter comprises: a sales counter (1), a camera (2) and an illuminating lamp (3);
the camera (2), the illuminating lamp (3) and the commodity (4) are all arranged in the sales counter (1), and the camera is arranged above the commodity (4) in the sales counter (1) and shoots the characteristics of the commodity and artificial characteristics on part of the commodity.
3. The artificial feature based article identification method according to claim 2, wherein the article positioning step:
positioning by using an artificial characteristic commodity training model according to the uploaded picture to obtain coordinates of all artificial characteristics in the picture, framing the artificial characteristics according to the coordinates of all artificial characteristics, intercepting the picture of the content in the frame according to the positioning frame, and outputting a positioning result and a screenshot;
and for the commodities without the adhered artificial characteristics, positioning the commodities by using the commodity training model with obvious characteristics, and outputting a positioning result.
4. The artificial feature based commodity identification method according to claim 3, wherein the commodity identification step:
for commodities without artificial features, identifying by using a commodity training model with obvious features according to the obtained positioning result to obtain a commodity identification result;
for the commodity pasted with the artificial features, decoding the obtained screenshot, wherein the decoding comprises the following steps:
performing opening operation and closing operation on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a first decoding result;
carrying out contrast enhancement on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a second decoding result;
carrying out distortion correction on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a third decoding result;
according to the obtained first, second and third decoding results, if two or more same results appear in the three decoding results of each screenshot, the decoding result is regarded as an artificial feature recognition result; if only one decoding result appears and the other two methods cannot decode, taking the decoding result as an artificial feature identification result;
judging whether the artificial characteristic identification result is empty: if the artificial characteristic identification result is empty, entering a full-image review step to continue execution; if the artificial characteristic identification result is not null, obtaining a commodity identification result and entering a commodity counting step for continuous execution;
judging whether the artificial feature recognition result is empty:
and if the following conditions occur, judging that the artificial feature recognition result is empty: all three methods cannot be decoded; one decoding result is null, and the other two decoding results are different; the three decoding results are all different;
the opening operation and the closing operation:
the opening and closing operation is a morphological processing method, and the operation of firstly corroding and then expanding is called as opening operation;
the closed operation refers to an operation of expansion followed by corrosion.
5. An artificial characteristic based article identification system, comprising:
an image acquisition module: shooting commodities in a sales counter, triggering image acquisition when a door of the sales counter is opened and closed, and uploading images acquired by the image acquisition;
a commodity positioning module: according to the uploaded picture, positioning the feature to be recognized by using the obtained feature training model to obtain a positioning result;
the commodity identification and clearing module: identifying and clearing according to the obtained positioning result and the characteristic to be identified;
further comprising:
a feature preparation module: making artificial characteristics to be attached to commodities with unobvious characteristics, and putting the commodities into a sales counter;
a model training module: carrying out neural network training on the commodity features of the commodities with obvious features to obtain a commodity training model with obvious features; for the commodities with the artificial characteristics, carrying out neural network training on the artificial characteristics of the commodities to obtain artificial characteristic commodity training models;
the feature training model comprises: the system comprises an obvious characteristic commodity training model and an artificial characteristic commodity training model;
the features to be identified include: commodity characteristics, artificial characteristics;
the commodity identification and clearing module comprises:
a positioning result judging module: judging whether the positioning result is empty: if the positioning result is null, the whole image review module is entered for continuous execution, and if the positioning result is not null, the commodity identification module is entered for continuous execution;
a commodity identification module: according to the obtained positioning result, identifying the commodity characteristics and the artificial characteristics to obtain a commodity identification result and an artificial characteristic identification result, if the artificial characteristic identification result is empty, obtaining the commodity identification result and calling a full-image review module, and if the artificial characteristic identification result is not empty, obtaining the commodity identification result and calling a commodity counting module; in the commodity counting module, according to a commodity identification result and an artificial characteristic identification result, comparing commodity information in a sales counter at two moments of opening and closing a door to obtain a final result of commodity identification;
the whole-image review module: performing distortion processing on the whole image, then decoding according to an industrial decoding standard, taking a decoding result as a final commodity identification result, and calling a commodity counting module;
the commodity counting module: according to the obtained commodity identification result, comparing the commodity information in the sales counter at the two moments of opening and closing the door to obtain the final result of commodity identification;
the feature preparation module:
the artificial features include: two-dimensional codes and dot matrix codes;
the sales counter comprises: a sales counter (1), a camera (2) and an illuminating lamp (3);
the camera (2), the illuminating lamp (3) and the commodity (4) are all arranged in the sales counter (1), and the camera is arranged above the commodity (4) in the sales counter (1) and is used for shooting the characteristics of the commodity and artificial characteristics on part of the commodity;
the commodity positioning module:
positioning by using an artificial characteristic commodity training model according to the uploaded picture to obtain coordinates of all artificial characteristics in the picture, framing the artificial characteristics according to the coordinates of all artificial characteristics, intercepting the picture of the content in the frame according to the positioning frame, and outputting a positioning result and a screenshot;
for commodities without artificial characteristics, positioning the commodities by using the commodity training model with obvious characteristics, and outputting a positioning result;
the commodity identification module:
for commodities without artificial features, identifying by using a commodity training model with obvious features according to the obtained positioning result to obtain a commodity identification result;
for the commodity pasted with the artificial features, decoding the obtained screenshot, wherein the decoding comprises the following steps:
performing opening operation and closing operation on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a first decoding result;
carrying out contrast enhancement on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a second decoding result;
carrying out distortion correction on the artificial features in the screenshot, and then decoding according to an industrial decoding standard corresponding to the artificial features to obtain a third decoding result;
according to the obtained first, second and third decoding results, if two or more same results appear in the three decoding results of each screenshot, the decoding result is regarded as an artificial feature recognition result; if only one decoding result appears and the other two methods cannot decode, taking the decoding result as an artificial feature identification result;
judging whether the artificial characteristic identification result is empty: if the artificial characteristic identification result is null, calling a full-image review module; if the artificial characteristic identification result is not null, obtaining a commodity identification result and calling a commodity counting module;
judging whether the artificial feature recognition result is empty:
and if the following conditions occur, judging that the artificial feature recognition result is empty: all three methods cannot be decoded; one decoding result is null, and the other two decoding results are different; the three decoding results are all different;
the opening operation and the closing operation:
the opening and closing operation is a morphological processing method, and the operation of firstly corroding and then expanding is called as opening operation;
the closed operation refers to an operation of expansion followed by corrosion.
6. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, performs the steps of the artificial feature based article identification method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910549890.4A CN110443118B (en) | 2019-06-24 | 2019-06-24 | Commodity identification method, system and medium based on artificial features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910549890.4A CN110443118B (en) | 2019-06-24 | 2019-06-24 | Commodity identification method, system and medium based on artificial features |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110443118A CN110443118A (en) | 2019-11-12 |
| CN110443118B true CN110443118B (en) | 2021-09-03 |
Family
ID=68428262
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910549890.4A Active CN110443118B (en) | 2019-06-24 | 2019-06-24 | Commodity identification method, system and medium based on artificial features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110443118B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111144417B (en) * | 2019-12-27 | 2023-08-01 | 创新奇智(重庆)科技有限公司 | Intelligent container small target detection method and detection system based on teacher and student network |
| CN111209911A (en) * | 2020-01-07 | 2020-05-29 | 创新奇智(合肥)科技有限公司 | Custom tag identification system and identification method based on semantic segmentation network |
| CN111340009A (en) * | 2020-05-15 | 2020-06-26 | 支付宝(杭州)信息技术有限公司 | Identification method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108920992A (en) * | 2018-08-08 | 2018-11-30 | 长沙理工大学 | A kind of positioning and recognition methods of the medical label bar code based on deep learning |
| CN109635705A (en) * | 2018-12-05 | 2019-04-16 | 上海交通大学 | A kind of commodity recognition method and device based on two dimensional code and deep learning |
| CN109816045A (en) * | 2019-02-11 | 2019-05-28 | 青岛海信智能商用***股份有限公司 | A kind of commodity recognition method and device |
| CN109919211A (en) * | 2019-02-26 | 2019-06-21 | 南京旷云科技有限公司 | Commodity recognition method, device, system and computer-readable medium |
-
2019
- 2019-06-24 CN CN201910549890.4A patent/CN110443118B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108920992A (en) * | 2018-08-08 | 2018-11-30 | 长沙理工大学 | A kind of positioning and recognition methods of the medical label bar code based on deep learning |
| CN109635705A (en) * | 2018-12-05 | 2019-04-16 | 上海交通大学 | A kind of commodity recognition method and device based on two dimensional code and deep learning |
| CN109816045A (en) * | 2019-02-11 | 2019-05-28 | 青岛海信智能商用***股份有限公司 | A kind of commodity recognition method and device |
| CN109919211A (en) * | 2019-02-26 | 2019-06-21 | 南京旷云科技有限公司 | Commodity recognition method, device, system and computer-readable medium |
Non-Patent Citations (1)
| Title |
|---|
| 商品识别结算***的设计与实现;张浩淼;《计算机软件及计算机应用》;20181231;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110443118A (en) | 2019-11-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110443118B (en) | Commodity identification method, system and medium based on artificial features | |
| US11182592B2 (en) | Target object recognition method and apparatus, storage medium, and electronic device | |
| CN109145901B (en) | Article identification method, apparatus, computer readable storage medium and computer device | |
| CN109508688B (en) | Skeleton-based behavior detection method, terminal equipment and computer storage medium | |
| CN106650662B (en) | Target object shielding detection method and device | |
| CN108154105B (en) | Underwater biological detection and identification method and device, server and terminal equipment | |
| CN109271847B (en) | Abnormity detection method, device and equipment in unmanned settlement scene | |
| US20190122072A1 (en) | Reverse neural network for object re-identification | |
| CN108846365B (en) | Detection method and device for fighting behavior in video, storage medium and processor | |
| US9471889B2 (en) | Video tracking based method for automatic sequencing of vehicles in drive-thru applications | |
| CN109727275B (en) | Object detection method, device, system and computer readable storage medium | |
| CN113111844B (en) | Operation posture evaluation method and device, local terminal and readable storage medium | |
| CN113962274B (en) | Abnormity identification method and device, electronic equipment and storage medium | |
| CN111626201A (en) | Commodity detection method and device and readable storage medium | |
| CN109791615A (en) | For detecting and tracking the method, target object tracking equipment and computer program product of target object | |
| CN112417955B (en) | Method and device for processing tour inspection video stream | |
| CN110570442A (en) | Contour detection method under complex background, terminal device and storage medium | |
| CN110909684A (en) | Working state checking system and method based on human body detection | |
| CN114511589A (en) | Human body tracking method and system | |
| CN111767754A (en) | Identification code identification method and device, electronic equipment and storage medium | |
| CN107862314B (en) | Code spraying identification method and device | |
| CN111242094B (en) | Commodity identification method, intelligent container and intelligent container system | |
| CN114037892A (en) | Method, device and equipment for target identification and readable storage medium | |
| MX2012001664A (en) | Recognition of objects. | |
| CN113743342B (en) | Method, system, terminal and storage medium for detecting assembly flow |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |