CN111428796B - General object detection method and system based on deep learning - Google Patents
General object detection method and system based on deep learning Download PDFInfo
- Publication number
- CN111428796B CN111428796B CN202010229590.0A CN202010229590A CN111428796B CN 111428796 B CN111428796 B CN 111428796B CN 202010229590 A CN202010229590 A CN 202010229590A CN 111428796 B CN111428796 B CN 111428796B
- Authority
- CN
- China
- Prior art keywords
- image
- article
- deep learning
- detected
- layer
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 38
- 238000013135 deep learning Methods 0.000 title claims abstract description 27
- 238000013136 deep learning model Methods 0.000 claims abstract description 33
- 238000002372 labelling Methods 0.000 claims abstract description 6
- 238000011176 pooling Methods 0.000 claims description 18
- 210000002569 neuron Anatomy 0.000 claims description 12
- 238000013527 convolutional neural network Methods 0.000 claims description 7
- 238000011478 gradient descent method Methods 0.000 claims description 6
- 239000000203 mixture Substances 0.000 claims description 2
- 238000000034 method Methods 0.000 abstract description 13
- 238000005286 illumination Methods 0.000 abstract description 3
- 239000003086 colorant Substances 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides a general object detection method based on deep learning. Collecting an image of an object to be detected; labeling boundaries according to the position information to obtain each article in the article image to be detected, distinguishing each article from the image background, and constructing a training data set; training by using the training data set and obtaining a deep learning model for detecting the position information of the article; and detecting all the articles in the image of the article to be detected by using the deep learning model to obtain the position information of each article in the image, and automatically marking each article in the image. Meanwhile, a general object detection system based on deep learning is provided. The invention has no limitation on the types of articles, has the characteristics of short training period, high running efficiency and the like, can quickly reach the expected training requirement and is practical to be put on line. Meanwhile, the method has good generalization performance, and when a depth image is adopted, the generalization performance is more outstanding, and the method has good robustness to different articles and different environments, such as weather conditions, illumination intensity and the like.
Description
Technical Field
The invention relates to the technical field of deep learning and computer vision, in particular to a general object detection method and system based on deep learning.
Background
The existing detection method based on deep learning generally adopts the following modes: after the traditional image of the object is acquired, a trained deep learning model is input, the type and the position of the object in the model output image are generally represented by rectangular frames, and the mode has the following obvious defects:
1. because the object type identification is needed, a large amount of image data of the specified object type is needed to be obtained in advance for training, the data acquisition is difficult, and the practical period of model training and online is long.
2. Since training and detection are performed only for specific several categories of articles, and the acquired data is limited, depth information is lacking, recognition errors are easily generated when the category is changed, or the category is unchanged but the feature is changed obviously, or the environmental conditions such as illumination and the like are changed. The generalization performance is not enough, and the robustness is not easy to guarantee.
3. The method is limited by the prior art, and the accuracy is reduced and the effect stability can not be ensured when various and large-quantity articles are detected.
No description or report of similar technology is found at present, and similar data at home and abroad are not collected.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a general object detection method and system based on deep learning, which have the characteristics of short training period, high running efficiency, strong generalization performance and the like, and can be rapidly and practically used on line.
The invention is realized by the following technical scheme.
According to one aspect of the present invention, there is provided a general item detection method based on deep learning, including:
collecting a traditional image of an object to be detected;
marking each article in the article image to be detected obtained by the boundary according to the position information, distinguishing each article from the image background, and constructing a training data set;
training by using the obtained training data set and obtaining a deep learning model for detecting the position information of all articles in the image;
and detecting all the articles in the image of the article to be detected by using the trained deep learning model to obtain the position information of each article in the image, and automatically marking each article in the image.
Preferably, the generalization performance is improved by replacing the conventional image with a depth image.
Preferably, the object to be detected is an object of the same kind or an object mixed by a plurality of kinds.
Preferably, the object to be detected refers to any object different from the background, the kind and the number of the object to be detected are not limited, and in the detection result, the kind of the object to be detected is not distinguished.
Preferably, the conventional image refers to an image representing a scene using a color space. The conventional images include, but are not limited to: RGB images, gray-scale images, etc. Wherein: RGB images are images in which a variety of colors are obtained by superimposing three color channels of red (R), green (G), and blue (B) on each other. A gray scale map refers to an image represented in gray scale, wherein: the logarithmic relationship between white and black is divided into several levels, called gray levels.
Preferably, the depth image refers to an image containing depth information, which is distance information from an image collector to points in the scene, which directly reflects the geometry of the visible surface in the scene. The depth image may be obtained by a depth camera or any other means including, but not limited to: RGBD image, single channel depth image of D channel only. The generalization performance of the method is improved by utilizing the depth information of the image.
Preferably, labeling each item in the image of the to-be-detected item according to the position information refers to labeling the boundary at the corresponding position according to the position of each item in the image, and distinguishing each item from the image background.
Preferably, the deep learning model adopts a convolutional neural network, and comprises: an input layer, a plurality of convolution layers, a plurality of pooling layers, a plurality of full connection layers and an output layer; wherein:
the input layer is used for inputting the object images with the same image size after the size is adjusted;
the convolution layer adopts a plurality of convolution kernels and is used for obtaining image features of the object image; when the characteristic learning is carried out on the convolution layer, the weight and the bias of the convolution kernel are adjusted by adopting a gradient descent method;
the pooling layer is used for carrying out mean pooling operation on the obtained image characteristics and reducing the data size of the image characteristics to one fourth of the original data size;
the full-connection layer is used for connecting the neurons of the current layer with the neurons of the previous layer;
the output layer is used for calculating classification and regression results.
Preferably, the automatically marking each article in the image means marking a boundary at a corresponding position according to the obtained position information, and distinguishing each article from the image background.
According to another aspect of the present invention, there is provided a general item detection system based on deep learning, including:
and an image acquisition module: collecting a traditional image of an object to be detected;
the deep learning model generation module: marking boundaries on the object images to be detected, which are obtained by the image acquisition module, according to the position information to obtain each object in the object images to be detected, distinguishing each object from the image background, and constructing a training data set; training by using the obtained training data set and obtaining a deep learning model for detecting the position information of all articles in the image;
an article detection module: and detecting all the articles in the image of the article to be detected by using the trained deep learning model obtained by the deep learning model generating module to obtain the position information of each article in the image, and automatically marking each article in the image.
Preferably, the object to be detected refers to any object different from the background, the kind and the number of the object to be detected are not limited, and in the detection result, the kind of the object to be detected is not distinguished.
Preferably, the deep learning model adopts a convolutional neural network, and comprises: an input layer, a plurality of convolution layers, a plurality of pooling layers, a plurality of full connection layers and an output layer; wherein:
the input layer is used for inputting the object images with the same image size after the size is adjusted;
the convolution layer adopts a plurality of convolution kernels and is used for obtaining image features of the object image; when the characteristic learning is carried out on the convolution layer, the weight and the bias of the convolution kernel are adjusted by adopting a gradient descent method;
the pooling layer is used for carrying out mean pooling operation on the obtained image characteristics and reducing the data size of the image characteristics to one fourth of the original data size;
the full-connection layer is used for connecting the neurons of the current layer with the neurons of the previous layer;
the output layer is used for calculating classification and regression results.
Preferably, replacing the traditional image of the object to be detected with a depth image of the object to be detected; wherein:
the traditional image refers to an image for expressing a scene by using a color space;
the depth image refers to an image containing depth information, wherein the depth information is distance information from an image collector to each point in a scene, and the information directly reflects the geometric shape of a visible surface in the scene.
Preferably, the object to be detected is an object of the same kind or a mixed object of a plurality of kinds different from the background. Compared with the prior art, the invention has at least one of the following beneficial effects:
1. the general object detection method and system based on deep learning provided by the invention have the advantages that the object types are not limited, the training data collection is convenient, the model training period is short, and the practical efficiency of online is high.
2. The general object detection method and system based on deep learning provided by the invention are not limited to specific objects of one or more types, and have little dependence on specific object characteristics, environmental conditions and the like, and have good generalization performance under different detection environments and robust results. After combining the depth information, the generalization performance can be further improved.
3. According to the method and the system for detecting the general articles based on the deep learning, provided by the invention, the difference between the articles and the background is emphasized by the model, the model is insensitive to the quantity and the type, and the stability of the result can be ensured when various large-quantity articles are detected.
4. According to the general object detection method and system based on deep learning, the object position result can be obtained and combined with any object classification method to obtain object type information, and function expansion is convenient.
Drawings
Other features, objects and advantages of the present invention will become more apparent upon reading of the detailed description of non-limiting embodiments, given with reference to the accompanying drawings in which:
fig. 1 is a flowchart of a general object detection method based on deep learning according to an embodiment of the present invention.
Fig. 2 is a schematic structural diagram of a general object detection system based on deep learning according to an embodiment of the present invention.
Detailed Description
The following describes embodiments of the present invention in detail: the embodiment is implemented on the premise of the technical scheme of the invention, and detailed implementation modes and specific operation processes are given. It should be noted that variations and modifications can be made by those skilled in the art without departing from the spirit of the invention, which falls within the scope of the invention.
The embodiment of the invention provides a general object detection method based on deep learning, which is shown in fig. 1 and comprises the following steps:
s1, collecting a traditional image of an object to be detected; further, the traditional image can be replaced by a depth image, so that the generalization performance is improved; the articles to be detected can be the same type of articles or various types of mixed articles, the articles to be detected refer to any articles different from the background, the types and the number of the articles are not limited, and in the detection result, the types of the articles to be detected are not distinguished.
S2, according to the mode of manually marking the boundary by the position information, obtaining each article in the image obtained in the S1, distinguishing each article from the image background, and constructing a training data set; here, the boundary is marked only by the positional information, and the article type is not marked.
And S3, training by using the training data set obtained in the step S2, and obtaining a deep learning model for detecting the position information of all the objects in the image.
S4, detecting all articles in the traditional image to be detected by using the trained deep learning model obtained in the S3 to obtain the position information of each article in the image, and automatically marking each article in the image; the article types are not distinguished here.
The article position result obtained by the article detection method can be combined with any article classification method in the follow-up to obtain article type information, and the type of the article is identified, so that the function expansion is convenient.
Further, in S1, the to-be-detected object refers to any object different from the background, the kind and the number of the to-be-detected object are not limited, and in the detection result, the kind of the to-be-detected object is not distinguished.
Further, in S1, the conventional image refers to an image that expresses a scene by using a color space, including but not limited to: RGB images, gray-scale images, etc. Wherein: RGB images are images in which a variety of colors are obtained by superimposing three color channels of red (R), green (G), and blue (B) on each other. A gray scale map refers to an image represented in gray scale, wherein: the logarithmic relationship between white and black is divided into several levels, called gray levels.
Further, in S1, the depth image is used as a training data set and an image to be detected, and the generalization performance of the method can be improved by using the depth information.
Further, in S1, the depth image refers to an image containing depth information, i.e. distance (depth) information from an image collector to points in the scene, which directly reflects the geometry of the visible surface in the scene. The image may be obtained by a depth camera or any other means. Including but not limited to: RGBD image, single channel depth image of D channel only.
Further, in S2, manually labeling each item in the image refers to labeling a rectangular frame or other boundaries with any shape at the corresponding positions according to the positions of each item in the image, and distinguishing each item from the image background.
Further, in S3, the deep learning model includes, but is not limited to: convolutional neural networks. The convolutional neural network comprises an input layer, a plurality of convolutional layers, a plurality of pooling layers, a plurality of full-connection layers and an output layer; wherein:
the input layer is used for inputting the object images with the same image size after the size is adjusted;
the convolution layer adopts a plurality of convolution kernels and is used for obtaining image features of the object image; when the characteristic learning is carried out on the convolution layer, the weight and the bias of the convolution kernel are adjusted by adopting a gradient descent method;
the pooling layer is used for carrying out mean pooling operation on the obtained image characteristics and reducing the data size of the image characteristics to one fourth of the original data size;
the full-connection layer is used for connecting the neurons of the current layer with the neurons of the previous layer;
the output layer is used for calculating classification and regression results;
further, in S4, the step of automatically marking each item in the image refers to marking a rectangular frame or other boundaries with any shape at the corresponding position according to the obtained position information, and distinguishing each item from the image background.
Based on the general object detection method based on the deep learning provided by the embodiment of the invention, the embodiment of the invention also provides a general object detection system based on the deep learning, and the system can be used for executing the method.
As shown in fig. 2, the system includes:
and an image acquisition module: collecting a traditional image of an object to be detected; further, the traditional image can be replaced by a depth image, so that the generalization performance is improved; the articles to be detected can be the same type of articles or various types of mixed articles, the articles to be detected refer to any articles different from the background, the types and the number of the articles are not limited, and in the detection result, the types of the articles to be detected are not distinguished.
A deep learning model generation module (simply referred to as a deep learning model module in fig. 2): marking boundaries on the object images to be detected, which are obtained by the image acquisition module, according to the position information to obtain each object in the object images to be detected, distinguishing each object from the image background, and constructing a training data set; training by using the obtained training data set and obtaining a deep learning model for detecting the position information of all articles in the image;
an article detection module: and detecting all the articles in the image of the article to be detected by using the trained deep learning model obtained by the deep learning model generating module to obtain the position information of each article in the image, and automatically marking each article in the image.
Further, the deep learning model adopts a convolutional neural network, and comprises the following steps: an input layer, a plurality of convolution layers, a plurality of pooling layers, a plurality of full connection layers and an output layer; wherein:
the input layer is used for inputting the object images with the same image size after the size is adjusted;
the convolution layer adopts a plurality of convolution kernels and is used for obtaining image features of the object image; when the characteristic learning is carried out on the convolution layer, the weight and the bias of the convolution kernel are adjusted by adopting a gradient descent method;
the pooling layer is used for carrying out mean pooling operation on the obtained image characteristics and reducing the data size of the image characteristics to one fourth of the original data size;
the full-connection layer is used for connecting the neurons of the current layer with the neurons of the previous layer;
the output layer is used for calculating classification and regression results.
Further, the traditional image of the object to be detected is replaced by a depth image; wherein:
the traditional image refers to an image for expressing a scene by using a color space;
the depth image refers to an image containing depth information, wherein the depth information is distance information from an image collector to each point in a scene, and the information directly reflects the geometric shape of a visible surface in the scene.
Further, the object to be detected is an object of the same kind or a mixed object of a plurality of kinds different from the background.
According to the general object detection method and system based on deep learning provided by the embodiment of the invention, the traditional image of the object to be detected is collected, and the object can be the same kind of object or a mixture of multiple kinds of objects; and manually marking each object in the image, marking the boundary only according to the position information, and not marking the type of the object. Distinguishing each article from the image background to construct a training data set; training and obtaining a deep learning model for detecting the position information of all articles in the image by using the obtained training data set; and detecting all the articles in the traditional image to be detected by using the trained deep learning model, and not distinguishing the types of the articles to obtain the position information of each article in the image, and marking each article in the image. In the optimization scheme, the depth image can be used for replacing the traditional image, so that the generalization performance is improved. The method and the system provided by the embodiment of the invention have the characteristics of no limitation on the article category, short training period, high running efficiency and the like, and can quickly reach the expected training requirement and be put on line practically. Meanwhile, the method has good generalization performance, and when a depth image is adopted, the generalization performance is more outstanding, and the method has good robustness to different articles and different environments, such as weather conditions, illumination intensity and the like.
The foregoing describes specific embodiments of the present invention. It is to be understood that the invention is not limited to the particular embodiments described above, and that various changes and modifications may be made by one skilled in the art within the scope of the claims without affecting the spirit of the invention.
Claims (7)
1. A general object detection method based on deep learning is characterized in that:
collecting a depth image of an object to be detected; the depth image is an image containing depth information, wherein the depth information is distance information from an image collector to each point in a scene, and the information directly reflects the geometric shape of a visible surface in the scene;
labeling boundaries according to the position information to obtain each article in the image of the article to be detected, distinguishing each article from the background of the image, and constructing a training data set;
training by using the obtained training data set and obtaining a deep learning model for detecting the position information of all articles in the image; the deep learning model learns differences between objects and backgrounds;
detecting all articles in the image of the article to be detected by using the trained deep learning model to obtain the position information of each article in the image, and automatically marking each article in the image;
each article in the image of the article to be detected, which is obtained by marking the boundary according to the position information, is marked with the boundary at the corresponding position according to the position of each article in the image, and each article is distinguished from the background of the image;
the object to be detected is any object different from the background, the type and the number of the object to be detected are not limited, and the type of the object to be detected is not distinguished in the detection result;
the automatic marking of each article in the image means that according to the obtained position information, a boundary is marked at the corresponding position, and each article is distinguished from the image background.
2. The general item detection method based on deep learning according to claim 1, wherein the item to be detected is the same kind of item or a mixed kind of items different from the background.
3. The general object detection method based on deep learning according to claim 1, wherein the conventional image refers to an image representing a scene using a color space.
4. The deep learning-based generic object detection method of claim 1, wherein the deep learning model employs a convolutional neural network, comprising: an input layer, a plurality of convolution layers, a plurality of pooling layers, a plurality of full connection layers and an output layer; wherein:
the input layer is used for inputting the object images with the same image size after the size is adjusted;
the convolution layer adopts a plurality of convolution kernels and is used for obtaining image features of the object image; when the characteristic learning is carried out on the convolution layer, the weight and the bias of the convolution kernel are adjusted by adopting a gradient descent method;
the pooling layer is used for carrying out mean pooling operation on the obtained image characteristics and reducing the data size of the image characteristics to one fourth of the original data size;
the full-connection layer is used for connecting the neurons of the current layer with the neurons of the previous layer;
the output layer is used for calculating classification and regression results.
5. A deep learning-based generic article detection system, comprising:
and an image acquisition module: collecting a depth image of an object to be detected; the depth image is an image containing depth information, wherein the depth information is distance information from an image collector to each point in a scene, and the information directly reflects the geometric shape of a visible surface in the scene;
the deep learning model generation module: marking boundaries on the object images to be detected, which are obtained by the image acquisition module, according to the position information to obtain each object in the object images to be detected, distinguishing each object from the image background, and constructing a training data set; training by using the obtained training data set and obtaining a deep learning model for detecting the position information of all articles in the image; the deep learning model learns differences between objects and backgrounds;
an article detection module: detecting all articles in the image of the article to be detected by using the trained deep learning model obtained by the deep learning model generating module to obtain the position information of each article in the image, and automatically marking each article in the image;
each object in the object image to be detected, which is obtained by marking the boundary according to the position information, in the deep learning model generating module refers to marking the boundary at the corresponding position according to the position of each object in the image, and distinguishing each object from the image background;
the object to be detected is any object different from the background, the type and the number of the object to be detected are not limited, and the type of the object to be detected is not distinguished in the detection result;
the automatic marking of each article in the image means that according to the obtained position information, a boundary is marked at the corresponding position, and each article is distinguished from the image background.
6. The deep learning based generic item detection system of claim 5, wherein the deep learning model employs a convolutional neural network comprising: an input layer, a plurality of convolution layers, a plurality of pooling layers, a plurality of full connection layers and an output layer; wherein:
the input layer is used for inputting the object images with the same image size after the size is adjusted;
the convolution layer adopts a plurality of convolution kernels and is used for obtaining image features of the object image; when the characteristic learning is carried out on the convolution layer, the weight and the bias of the convolution kernel are adjusted by adopting a gradient descent method;
the pooling layer is used for carrying out mean pooling operation on the obtained image characteristics and reducing the data size of the image characteristics to one fourth of the original data size;
the full-connection layer is used for connecting the neurons of the current layer with the neurons of the previous layer;
the output layer is used for calculating classification and regression results.
7. The deep learning based universal item detection system of claim 5, wherein the item to be detected is the same kind of item or a mixture of kinds of items other than background.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010229590.0A CN111428796B (en) | 2020-03-27 | 2020-03-27 | General object detection method and system based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010229590.0A CN111428796B (en) | 2020-03-27 | 2020-03-27 | General object detection method and system based on deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111428796A CN111428796A (en) | 2020-07-17 |
CN111428796B true CN111428796B (en) | 2024-04-02 |
Family
ID=71549059
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010229590.0A Active CN111428796B (en) | 2020-03-27 | 2020-03-27 | General object detection method and system based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111428796B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112007352B (en) * | 2020-08-20 | 2022-01-04 | 北方民族大学 | Volleyball boundary processing device and method based on deep learning and photoelectric detection |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2016165082A1 (en) * | 2015-04-15 | 2016-10-20 | 中国科学院自动化研究所 | Image stego-detection method based on deep learning |
CN107067015A (en) * | 2016-10-27 | 2017-08-18 | 武汉烽火众智数字技术有限责任公司 | A kind of vehicle checking method and device based on multiple features deep learning |
CN107392896A (en) * | 2017-07-14 | 2017-11-24 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of Wood Defects Testing method and system based on deep learning |
CN109447071A (en) * | 2018-11-01 | 2019-03-08 | 博微太赫兹信息科技有限公司 | A kind of mm-wave imaging dangerous goods detection method based on FPGA and deep learning |
CN109766920A (en) * | 2018-12-18 | 2019-05-17 | 任飞翔 | Article characteristics Model Calculating Method and device based on deep learning |
CN109800864A (en) * | 2019-01-18 | 2019-05-24 | 中山大学 | A kind of robot Active Learning Method based on image input |
CN110472544A (en) * | 2019-08-05 | 2019-11-19 | 上海英迈吉东影图像设备有限公司 | A kind of training method and system of article identification model |
CN110503037A (en) * | 2019-08-22 | 2019-11-26 | 三星电子(中国)研发中心 | A kind of method and system of the positioning object in region |
-
2020
- 2020-03-27 CN CN202010229590.0A patent/CN111428796B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2016165082A1 (en) * | 2015-04-15 | 2016-10-20 | 中国科学院自动化研究所 | Image stego-detection method based on deep learning |
CN107067015A (en) * | 2016-10-27 | 2017-08-18 | 武汉烽火众智数字技术有限责任公司 | A kind of vehicle checking method and device based on multiple features deep learning |
CN107392896A (en) * | 2017-07-14 | 2017-11-24 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of Wood Defects Testing method and system based on deep learning |
CN109447071A (en) * | 2018-11-01 | 2019-03-08 | 博微太赫兹信息科技有限公司 | A kind of mm-wave imaging dangerous goods detection method based on FPGA and deep learning |
CN109766920A (en) * | 2018-12-18 | 2019-05-17 | 任飞翔 | Article characteristics Model Calculating Method and device based on deep learning |
CN109800864A (en) * | 2019-01-18 | 2019-05-24 | 中山大学 | A kind of robot Active Learning Method based on image input |
CN110472544A (en) * | 2019-08-05 | 2019-11-19 | 上海英迈吉东影图像设备有限公司 | A kind of training method and system of article identification model |
CN110503037A (en) * | 2019-08-22 | 2019-11-26 | 三星电子(中国)研发中心 | A kind of method and system of the positioning object in region |
Also Published As
Publication number | Publication date |
---|---|
CN111428796A (en) | 2020-07-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111612763B (en) | Mobile phone screen defect detection method, device and system, computer equipment and medium | |
CN109829914B (en) | Method and device for detecting product defects | |
Recky et al. | Windows detection using k-means in cie-lab color space | |
CN108921120B (en) | Cigarette identification method suitable for wide retail scene | |
CN106251333B (en) | Element reverse detection method and system | |
CN111242896A (en) | Color printing label defect detection and quality rating method | |
CN106919910B (en) | Traffic sign identification method based on HOG-CTH combined features | |
CN108564631A (en) | Car light light guide acetes chinensis method, apparatus and computer readable storage medium | |
CN112529901B (en) | Crack identification method in complex environment | |
WO2018010386A1 (en) | Method and system for component inversion testing | |
CN111967313A (en) | Unmanned aerial vehicle image annotation method assisted by deep learning target detection algorithm | |
Zhang et al. | Application research of YOLO v2 combined with color identification | |
CN105825168A (en) | Golden snub-nosed monkey face detection and tracking algorithm based on S-TLD | |
CN102676657A (en) | Sequencing image recognition system and sequencing image recognition method | |
CN111681220B (en) | Method, device, system and storage medium for constructing defect detection model | |
CN111461036A (en) | Real-time pedestrian detection method using background modeling enhanced data | |
CN113888754A (en) | Vehicle multi-attribute identification method based on radar vision fusion | |
CN111428796B (en) | General object detection method and system based on deep learning | |
CN110263608A (en) | Electronic component automatic identifying method based on image feature space variable threshold value metric | |
CN116740758A (en) | Bird image recognition method and system for preventing misjudgment | |
CN114627461A (en) | Method and system for high-precision identification of water gauge data based on artificial intelligence | |
CN113657339A (en) | Instrument pointer counting and reading method and medium based on machine vision | |
CN111414855B (en) | Telegraph pole sign target detection and identification method based on end-to-end regression model | |
CN106960188B (en) | Weather image classification method and device | |
Surinwarangkoon et al. | Traffic sign recognition system for roadside images in poor condition |
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 |