CN116229388B - Method, system and equipment for detecting motor car foreign matters based on target detection network - Google Patents
Method, system and equipment for detecting motor car foreign matters based on target detection network Download PDFInfo
- Publication number
- CN116229388B CN116229388B CN202310310547.0A CN202310310547A CN116229388B CN 116229388 B CN116229388 B CN 116229388B CN 202310310547 A CN202310310547 A CN 202310310547A CN 116229388 B CN116229388 B CN 116229388B
- Authority
- CN
- China
- Prior art keywords
- target detection
- detection model
- foreign matter
- motor car
- foreign
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
-
- 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
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Control Of Electric Motors In General (AREA)
- Image Analysis (AREA)
Abstract
The application discloses a motor car foreign matter detection method, system and equipment based on a target detection network, and relates to a motor car foreign matter detection method, system and equipment. The application aims to solve the problems that the existing method has a good detection effect on known foreign matter types, but the detection effect is not good for the foreign matter types which are not in the data set. The motor car foreign matter detection method based on the target detection network comprises the following specific processes: step one, constructing a training data set; training a target detection model based on the training data set constructed in the first step to obtain a trained target detection model; the loss function of the target detection model is that a binary cross entropy loss function of the Yolov5 target detection model is replaced by a single-classification loss function; and step three, detecting the image to be detected based on the trained target detection model, and judging whether foreign matters exist or not. The application is used in the field of motor car foreign matter detection.
Description
Technical Field
The application relates to a method, a system and equipment for detecting foreign matters of a motor car.
Background
Foreign matter faults are common fault types of motor cars. A method of detecting foreign matter in an image is generally to locate the position of the foreign matter in a pattern and obtain the kind of the foreign matter using a target detection network. The target detection network needs to train by using images with foreign matters, and the method has good detection effect on known foreign matters, but has poor detection effect on foreign matters which are not in the data set. The types of faults of the motor car foreign matters are variable, all fault forms are difficult to collect, and missed detection can be caused by using a target detection network.
Disclosure of Invention
The application aims to solve the problems that the existing method has a good detection effect on known foreign matter types, but the detection effect is not good for the foreign matter types which are not in a data set, and provides a motor car foreign matter detection method, a motor car foreign matter detection system and motor car foreign matter detection equipment based on a target detection network.
The motor car foreign matter detection method based on the target detection network comprises the following specific processes:
step one, constructing a training data set;
training a target detection model based on the training data set constructed in the first step to obtain a trained target detection model;
the loss function of the target detection model is that a binary cross entropy loss function of the Yolov5 target detection model is replaced by a single-classification loss function;
the single class loss function is defined as follows:
C=∑a×(1-y)×max{0,||x-c|| 2 -R 2 }+y×max{0,(R+d) 2 -x-c 2 }
wherein: x is a vector output by the target detection model; y is a label, and when y is equal to 1, the label represents a foreground class, and when y is equal to 0, the label represents a background class; c is the center of the hyper sphere, namely the average position of the background class, R is the radius of the hyper sphere, and d is a constant, so that the degree of distinction between the background class and the foreground class can be increased; a is a constant for adjusting the weight of the background and foreground in loss; i 2 Is the square of the Euclidean distance;
and step three, detecting the image to be detected based on the trained target detection model, and judging whether foreign matters exist or not.
Preferably, in step one, a training data set is constructed; the specific process is as follows:
collecting a normal image of a motor car as a background image;
collecting a motor car foreign matter image, and removing the background in the motor car foreign matter image through PS to obtain a foreign matter material;
randomly rotating, zooming, brightness and contrast ratio conversion is carried out on the foreign material picture;
randomly pasting the transformed foreign matter material into a background image to generate a picture with foreign matters, and generating labeling frame information of the foreign matters;
and taking the generated picture with the foreign matters and the label frame information as a training data set.
Preferably, the target detection model in the second step is a Yolov5 target detection model.
Preferably, training the target detection model based on the training data set constructed in the first step to obtain a trained target detection model; the specific process is as follows:
the input of the target detection model is a picture with foreign matters, the output of the target detection model is a labeling frame and a category, the category label of the foreign matters is a foreground, the value is 1, the category label of the non-foreign matters is a background, and the value is 0;
the loss function of the target detection model is that a binary cross entropy loss function of the Yolov5 target detection model is replaced by a single-classification loss function;
and obtaining a trained target detection model.
Preferably, the object detection model employs sigmoid as the activation function.
Preferably, the target detection model judges whether the target detection model belongs to the foreground or the background according to the distance between the output class vector and the center of the super ball; the specific process is as follows:
if the distance between the output vector of the category and the center of the super ball is larger than the radius R of the super ball, the method belongs to the prospect;
if the distance between the output vector of the category and the center of the hyper-sphere is smaller than or equal to the radius R of the hyper-sphere, the vector belongs to the background.
Preferably, in the third step, the image to be detected is detected based on the trained target detection model, and whether foreign matters exist or not is judged; the specific process is as follows:
inputting the image to be detected into a trained target detection model, judging the distance between the class vector output by the target detection model and the center c of the hypersphere, if the distance is larger than the radius R of the hypersphere, considering that foreign matters exist in the corresponding detection frame, and obtaining the position of the foreign matters through position regression.
The motor car foreign matter detection system based on the target detection network is a program for executing a motor car foreign matter detection method based on the target detection network.
A storage medium having stored therein at least one instruction that is loaded by a processor and that operates a motor vehicle foreign object detection system based on a target detection network.
The device comprises a processor and a memory, wherein at least one instruction is stored in the memory, and the processor loads and runs the motor vehicle foreign matter detection system based on the target detection network.
The application has the beneficial effects that:
in order to better detect the types of the foreign matters which are not in the data set, the foreign matter materials are used for random pasting and labeling, so that abnormal data can be generated more quickly, and the requirement on fault data collection is reduced; the present application replaces the binary cross entropy loss function in the target detection network with a single classification loss function. After the single-classification loss function is used, the target detection network can better distinguish the background and the foreign matters in the image, the target detection network takes the normal image as the background, the object with low similarity with the normal image can be detected as the foreign matters, and the detection effect on the types of the foreign matters which are not found is better.
Drawings
FIG. 1 is a flow chart of the present application.
Detailed Description
It should be noted that, in particular, the various embodiments of the present disclosure may be combined with each other without conflict.
The first embodiment is as follows: referring to fig. 1, the specific procedure of the method for detecting a foreign object in a motor vehicle based on a target detection network according to the present embodiment is as follows:
step one, constructing a motor car image training data set;
training a target detection model based on the training data set constructed in the first step to obtain a trained target detection model;
the loss function of the target detection model is that a binary cross entropy loss function of the Yolov5 target detection model is replaced by a single-classification loss function;
the single class loss function is defined as follows:
C=∑a×(1-y)×max{0,||x-c|| 2 -R 2 }+y×max{0,(R+d) 2 -x-c 2 }
wherein: x is a vector output by the target detection model; y is a label, y is equal to 1A foreground class (foreign object in the foreign object image) and a background class (non-foreign object region in the foreign object image) when y is equal to 0; c is the center of the hyper-sphere, namely the average position of the background class, R is the radius of the hyper-sphere (trainable variable), and d is a constant, so that the degree of distinction between the background class and the foreground class can be increased; a is a constant for adjusting the weight of the background and foreground in loss; i 2 Is the square of the Euclidean distance;
the loss function can concentrate the background class in the super-sphere, and the foreground class moves outside the super-sphere;
and step three, detecting the image to be detected based on the trained target detection model, and judging whether foreign matters exist or not.
The second embodiment is as follows: the first embodiment is different from the first embodiment in that a training data set is constructed in the first step; the specific process is as follows:
collecting a normal image of a motor car as a background image;
collecting a motor car foreign matter image, and removing the background in the motor car foreign matter image through PS to obtain a foreign matter material;
randomly rotating, zooming, brightness and contrast ratio conversion is carried out on the foreign material picture;
randomly pasting the transformed foreign matter material into a background image to generate a picture with foreign matters, and generating labeling frame information of the foreign matters;
and taking the generated picture with the foreign matters and the label frame information as a training data set.
Other steps and parameters are the same as in the first embodiment.
And a third specific embodiment: the difference between the present embodiment and the first or second embodiment is that the target detection model in the second step is a Yolov5 target detection model.
Other steps and parameters are the same as in the first or second embodiment.
The specific embodiment IV is as follows: the difference between the embodiment and the first to third embodiments is that, in the second step, the target detection model is trained based on the training data set constructed in the first step, and a trained target detection model is obtained; the specific process is as follows:
the input of the target detection model is a picture with foreign matters, the output of the target detection model is a labeling frame and a category, the category label of the foreign matters is a foreground, the value is 1, the category label of the non-foreign matters is a background, and the value is 0;
the loss function of the target detection model is that a binary cross entropy loss function of the Yolov5 target detection model is replaced by a single-classification loss function;
the classification loss function of Yolov5 is modified.
The classification loss function of original Yolov5 is a binary cross entropy loss function. And judging the probability that the object belongs to each category.
The rest training process is consistent with the original Yolov5 model.
And obtaining a trained target detection model.
Other steps and parameters are the same as in one to three embodiments.
Fifth embodiment: this embodiment differs from the first to fourth embodiments in that the object detection model uses sigmoid as an activation function.
Other steps and parameters are the same as in one to three embodiments.
Specific embodiment six: the difference between the embodiment and one to fifth embodiments is that the target detection model judges whether the target detection model belongs to the foreground or the background according to the distance between the output class vector and the center of the super ball; the specific process is as follows:
if the distance between the output vector of the category and the center of the super ball is larger than the radius R of the super ball, the method belongs to the prospect;
if the distance between the output vector of the category and the center of the hyper-sphere is smaller than or equal to the radius R of the hyper-sphere, the vector belongs to the background.
Other steps and parameters are the same as in one of the first to fifth embodiments.
Seventh embodiment: the difference between the embodiment and one of the first to sixth embodiments is that, in the third step, the image to be detected is detected based on the trained target detection model, and whether the foreign matter exists is judged; the specific process is as follows:
inputting the image to be detected into a trained target detection model, judging the distance between the class vector output by the target detection model and the center c of the super ball, if the distance is larger than the radius R of the super ball, considering that foreign matters exist in the corresponding detection frame, and obtaining the accurate position of the foreign matters through position regression.
If the foreign matter type training set in the image is not available, but the difference between the foreign matter and the background is large, the distance between the detection vector output by the network and the center of the super ball is still large, so that the foreign matter is detected.
Other steps and parameters are the same as in one of the first to sixth embodiments.
Eighth embodiment: the motor vehicle foreign matter detection system based on the target detection network according to the present embodiment is a program for executing a motor vehicle foreign matter detection method based on the target detection network.
Detailed description nine: the present embodiment is a storage medium in which at least one instruction is stored, the at least one instruction being loaded by a processor and operating a motor car foreign matter detection system based on a target detection network.
It should be understood that any method, including those described herein, may be provided as a computer program product, software, or computerized method, which may include a non-transitory machine-readable medium having stored thereon instructions, which may be used to program a computer system, or other electronic device. The storage medium may include, but is not limited to, magnetic storage media, optical storage media; the magneto-optical storage medium includes: read only memory ROM, random access memory RAM, erasable programmable memory (e.g., EPROM and EEPROM), and flash memory layers; or other type of medium suitable for storing electronic instructions.
Detailed description ten: the embodiment is a motor car foreign matter detection device based on a target detection network, the device comprises a processor and a memory, at least one instruction is stored in the memory, and the at least one instruction is loaded by the processor and runs a motor car foreign matter detection system based on the target detection network.
The present application is capable of other and further embodiments and its several details are capable of modification and variation in light of the present application, as will be apparent to those skilled in the art, without departing from the spirit and scope of the application as defined in the appended claims.
Claims (7)
1. The motor car foreign matter detection method based on the target detection network is characterized by comprising the following steps of: the method comprises the following specific processes:
step one, constructing a training data set;
training a target detection model based on the training data set constructed in the first step to obtain a trained target detection model;
the loss function of the target detection model is that a binary cross entropy loss function of the Yolov5 target detection model is replaced by a single-classification loss function;
the single class loss function is defined as follows:
C=∑a×(1-y)×max{0,||x-c|| 2 -R 2 }+y×max{0,(R+d) 2 -||x-c|| 2 }
wherein: x is a vector output by the target detection model; y is a label, and when y is equal to 1, the label represents a foreground class, and when y is equal to 0, the label represents a background class; c is the center of the hyper sphere, namely the average position of the background class, R is the radius of the hyper sphere, and d is a constant, so that the degree of distinction between the background class and the foreground class can be increased; a is a constant for adjusting the weight of the background and foreground in loss; i 2 Is the square of the Euclidean distance;
detecting an image to be detected based on a trained target detection model, and judging whether foreign matters exist or not;
constructing a training data set in the first step; the specific process is as follows:
collecting a normal image of a motor car as a background image;
collecting a motor car foreign matter image, and removing the background in the motor car foreign matter image through PS to obtain a foreign matter material;
randomly rotating, zooming, brightness and contrast ratio conversion is carried out on the foreign material picture;
randomly pasting the transformed foreign matter material into a background image to generate a picture with foreign matters, and generating labeling frame information of the foreign matters;
taking the generated picture with the foreign matters and the label frame information as a training data set;
the target detection model in the second step is a Yolov5 target detection model;
training a target detection model based on the training data set constructed in the first step to obtain a trained target detection model; the specific process is as follows:
the input of the target detection model is a picture with foreign matters, the output of the target detection model is a labeling frame and a category, the category label of the foreign matters is a foreground, the value is 1, the category label of the non-foreign matters is a background, and the value is 0;
the loss function of the target detection model is that a binary cross entropy loss function of the Yolov5 target detection model is replaced by a single-classification loss function;
and obtaining a trained target detection model.
2. The method for detecting a foreign object on a motor car based on a target detection network according to claim 1, wherein: the target detection model adopts sigmoid as an activation function.
3. The method for detecting a foreign object on a motor car based on a target detection network according to claim 2, wherein: the target detection model judges whether the target detection model belongs to the foreground or the background according to the distance between the output class vector and the center of the super ball; the specific process is as follows:
if the distance between the output vector of the category and the center of the super ball is larger than the radius R of the super ball, the method belongs to the prospect;
if the distance between the output vector of the category and the center of the hyper-sphere is smaller than or equal to the radius R of the hyper-sphere, the vector belongs to the background.
4. A motor car foreign matter detection method based on a target detection network according to claim 3, characterized in that: detecting an image to be detected based on a trained target detection model, and judging whether foreign matters exist or not; the specific process is as follows:
inputting the image to be detected into a trained target detection model, judging the distance between the class vector output by the target detection model and the center c of the super ball, if the distance is larger than the radius R of the super ball, considering that foreign matters exist in the corresponding detection frame, and obtaining the position of the foreign matters through position regression.
5. Motor car foreign matter detecting system based on target detection network, its characterized in that: the object detection network-based motor vehicle foreign matter detection system is a program for executing the object detection network-based motor vehicle foreign matter detection method according to one of claims 1 to 4.
6. A storage medium having stored therein at least one instruction loaded by a processor and running the object detection network-based motor car foreign object detection system of claim 5.
7. A motor vehicle foreign object detection apparatus based on a target detection network, characterized in that the apparatus comprises a processor and a memory, wherein at least one instruction is stored in the memory, and wherein the at least one instruction is loaded by the processor and operates the motor vehicle foreign object detection system based on the target detection network as claimed in claim 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310310547.0A CN116229388B (en) | 2023-03-27 | 2023-03-27 | Method, system and equipment for detecting motor car foreign matters based on target detection network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310310547.0A CN116229388B (en) | 2023-03-27 | 2023-03-27 | Method, system and equipment for detecting motor car foreign matters based on target detection network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116229388A CN116229388A (en) | 2023-06-06 |
CN116229388B true CN116229388B (en) | 2023-09-12 |
Family
ID=86580647
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310310547.0A Active CN116229388B (en) | 2023-03-27 | 2023-03-27 | Method, system and equipment for detecting motor car foreign matters based on target detection network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116229388B (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110399816A (en) * | 2019-07-15 | 2019-11-01 | 广西大学 | A kind of high-speed train bottom foreign matter detecting method based on Faster R-CNN |
CN111368874A (en) * | 2020-01-23 | 2020-07-03 | 天津大学 | Image category incremental learning method based on single classification technology |
WO2020199834A1 (en) * | 2019-04-03 | 2020-10-08 | 腾讯科技(深圳)有限公司 | Object detection method and apparatus, and network device and storage medium |
WO2021233017A1 (en) * | 2020-05-18 | 2021-11-25 | 腾讯科技(深圳)有限公司 | Image processing method and apparatus, and device and computer-readable storage medium |
CN113793327A (en) * | 2021-09-18 | 2021-12-14 | 北京中科智眼科技有限公司 | High-speed rail foreign matter detection method based on token |
CN113903009A (en) * | 2021-12-10 | 2022-01-07 | 华东交通大学 | Railway foreign matter detection method and system based on improved YOLOv3 network |
CN114926456A (en) * | 2022-06-14 | 2022-08-19 | 华南农业大学 | Rail foreign matter detection method based on semi-automatic labeling and improved deep learning |
-
2023
- 2023-03-27 CN CN202310310547.0A patent/CN116229388B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020199834A1 (en) * | 2019-04-03 | 2020-10-08 | 腾讯科技(深圳)有限公司 | Object detection method and apparatus, and network device and storage medium |
CN110399816A (en) * | 2019-07-15 | 2019-11-01 | 广西大学 | A kind of high-speed train bottom foreign matter detecting method based on Faster R-CNN |
CN111368874A (en) * | 2020-01-23 | 2020-07-03 | 天津大学 | Image category incremental learning method based on single classification technology |
WO2021233017A1 (en) * | 2020-05-18 | 2021-11-25 | 腾讯科技(深圳)有限公司 | Image processing method and apparatus, and device and computer-readable storage medium |
CN113793327A (en) * | 2021-09-18 | 2021-12-14 | 北京中科智眼科技有限公司 | High-speed rail foreign matter detection method based on token |
CN113903009A (en) * | 2021-12-10 | 2022-01-07 | 华东交通大学 | Railway foreign matter detection method and system based on improved YOLOv3 network |
CN114926456A (en) * | 2022-06-14 | 2022-08-19 | 华南农业大学 | Rail foreign matter detection method based on semi-automatic labeling and improved deep learning |
Also Published As
Publication number | Publication date |
---|---|
CN116229388A (en) | 2023-06-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhang et al. | Machinery fault diagnosis with imbalanced data using deep generative adversarial networks | |
EP3379491B1 (en) | Surface defect detection | |
Xiao et al. | Surface defect detection using image pyramid | |
CN111275660B (en) | Flat panel display defect detection method and device | |
CN111968095B (en) | Product surface defect detection method, system, device and medium | |
CN112561910A (en) | Industrial surface defect detection method based on multi-scale feature fusion | |
CN115937655B (en) | Multi-order feature interaction target detection model, construction method, device and application thereof | |
CN112763214A (en) | Rolling bearing fault diagnosis method based on multi-label zero-sample learning | |
CN112766110A (en) | Training method of object defect recognition model, object defect recognition method and device | |
US20220222799A1 (en) | Method for detecting defect in products and electronic device using method | |
CN114708437B (en) | Training method of target detection model, target detection method, device and medium | |
CN117115412A (en) | Small target detection method based on weighted score label distribution | |
CN113781483B (en) | Industrial product appearance defect detection method and device | |
Shanthakumari et al. | Mask RCNN and Tesseract OCR for vehicle plate character recognition | |
CN116958561A (en) | Method, apparatus and storage medium for detecting abnormal object | |
CN116883390B (en) | Fuzzy-resistant semi-supervised defect detection method, device and storage medium | |
CN112560863B (en) | Method, system, device and medium for detecting ground cleanliness of garbage recycling station | |
CN116229388B (en) | Method, system and equipment for detecting motor car foreign matters based on target detection network | |
CN112836724A (en) | Object defect recognition model training method and device, electronic equipment and storage medium | |
Ibrahem et al. | Weakly supervised traffic sign detection in real time using single CNN architecture for multiple purposes | |
Kuhn et al. | Better look twice-improving visual scene perception using a two-stage approach | |
CN111814922B (en) | Video clip content matching method based on deep learning | |
CN114581722A (en) | Two-stage multi-classification industrial image defect detection method based on twin residual error network | |
Zhu et al. | Multi-size object detection assisting fault diagnosis of power systems based on improved cascaded faster R-CNNs | |
CN114298137A (en) | Tiny target detection system based on countermeasure generation network |
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 |