CN111680729A - Airport passenger security inspection dangerous goods real-time high-recognition rate method based on X-ray machine image - Google Patents
Airport passenger security inspection dangerous goods real-time high-recognition rate method based on X-ray machine image Download PDFInfo
- Publication number
- CN111680729A CN111680729A CN202010480894.4A CN202010480894A CN111680729A CN 111680729 A CN111680729 A CN 111680729A CN 202010480894 A CN202010480894 A CN 202010480894A CN 111680729 A CN111680729 A CN 111680729A
- Authority
- CN
- China
- Prior art keywords
- training
- image
- neural network
- dangerous goods
- model
- 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.)
- Pending
Links
Images
Classifications
-
- 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
- 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
- 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
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
Abstract
The invention provides a method for identifying dangerous goods of airport passenger security check in a high-speed and real-time manner based on an X-ray machine image, which comprises the following steps: constructing a dangerous article shape characteristic sample library and constructing a dangerous article color model sample; classifying and labeling the image data of the shape sample of the dangerous goods, and dividing the classified and labeled data into a training set, a verification set and a test set; establishing a deep learning neural network model; training a neural network model, and determining network model parameters; testing the trained neural network through a test set to set a type discrimination detection threshold and a positioning threshold of the dangerous goods; and combining the dangerous goods color model with a deep learning neural network to judge the type of the dangerous goods. The invention combines the contour, color information and local characteristics of different dangerous goods images to make the training more targeted and flexible, thereby effectively improving the detection accuracy of the system.
Description
Technical Field
The invention relates to the technical field of security check artificial intelligence identification, in particular to a method for realizing high identification rate of airport passenger security check dangerous goods in real time based on an X-ray machine image.
Background
Security inspection devices are continuously developed by many company organizations in the world, and among them, X-ray security inspection devices are widely used. In order to ensure the safety of airport passengers, dangerous articles such as knife guns, flammable and explosive articles and the like in luggage packages can be accurately identified during security inspection, and the probability of danger occurrence is reduced.
In order to improve the efficiency and accuracy of security check identification, reduce manual detection intervention of professional trainers and avoid false detection of missed detection, it is necessary to adopt a deep learning algorithm based on artificial intelligence and other auxiliary means to realize automatic identification of security check luggage dangerous goods, promote quality improvement and efficiency enhancement of security check services and improve the capability of civil aviation safety guarantee.
Disclosure of Invention
In order to solve the problems, the method for identifying the airport passenger security inspection dangerous goods in high real time based on the X-ray machine image is provided.
The object of the invention is achieved in the following way:
an airport passenger security inspection dangerous goods real-time high-identification-rate method based on X-ray machine images comprises the following steps:
s1: constructing a dangerous article shape characteristic sample library and constructing a dangerous article color model sample;
s2: classifying and labeling the image data of the shape sample of the dangerous goods, and dividing the classified and labeled data into a training set, a verification set and a test set;
s3: establishing a deep learning neural network model;
s4: training a neural network model, and determining network model parameters; inputting the training set serving as a sample into a neural network model for training, inputting the verification set serving as a sample into the neural network model, and verifying the accuracy of the neural network model; initial values of weight parameters and deviation parameters of the neural network model are given, and the training speed is improved; secondly, adjusting a network architecture, judging the effectiveness of training through the variation trend of the minimum variance sum of a training set and a verification set, if the loss of the training set continuously decreases, the loss of the verification set also continuously decreases, which indicates that the neural network continues to learn normally, and if the loss of the training set continuously decreases, the loss of the verification set tends to be unchanged, which indicates that the network is overfitting, so that the training can be stopped, and determining the weight parameters and the deviation parameters of the neural network;
s5: testing the trained neural network through a test set to set a type discrimination detection threshold and a positioning threshold of the dangerous goods;
s6: and (3) combining the dangerous article color model with a deep learning neural network to judge the type of the dangerous article: firstly, judging through a color model, and detecting by a deep learning method if no problem exists; the specific determination method of the color model is as follows: by converting the collected RGB image into an HSI image and combining threshold judgment of two color models, when the R channel numerical value and the brightness are greater than the set threshold and the image area is greater than the set threshold, an alarm is directly given, or when the image threshold of the RGB three channels is less than the set threshold, an alarm is given.
The dangerous article shape characteristic sample library is obtained by purchasing a dangerous article sample library, and a sample image library is constructed in a targeted manner, so that samples can be acquired from multiple visual angles.
The dangerous goods color model separates the collected X-ray image into an RGB three-channel color model, converts the RGB three-channel color model into an HIS color model, and takes the dangerous goods color model component relation of the image as a criterion for identifying dangerous goods; the method specifically comprises the following steps: firstly, an acquired three-channel X-ray image is separated into R, G, B three-channel color model images, then the three-channel color model images are converted into HIS color model images according to the images of the RGB three-channel color model, a saturation threshold value in the image HIS model is assumed to be ST, a red component threshold value in the RGB color model is assumed to be RT, and the images meeting the conditions of R > G > B and S > = (255-R) × ST/RT and continuous areas of more than 50 pixels are combined to serve as criteria of a dangerous goods color model.
And the dangerous goods color sample is collected by adopting an X-ray machine to collect images to establish an image sample library, and the image sample library contains a certain amount of images of the target object.
The classification in S2 is based on whether there is an obvious contour distinguishing feature, and if the contour can be distinguished from the outline, the entire contour is used for distinguishing; if the outline can not be distinguished, the outline of the part is used for distinguishing; and the classification labeling in the S2 adopts a classification labeling method of multi-key point characteristics.
In S4, if the training of the network is normal and the loss of the training set and the verification set is less than 5, the network learning may be stopped, and if the loss of the training set and the loss of the verification set are both less than the set threshold and the number of times of non-convergence is greater than the set threshold, the training of the network model may also be stopped.
The deep learning neural network model in S3 adopts a Darnet53 neural network as a backbone training network.
The invention has the beneficial effects that: compared with the prior art, the method and the system have the advantages that the training is more targeted and flexible according to the combination of the contours, the color information and the local features of different dangerous article images, and the detection accuracy of the system is effectively improved.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The technical solutions in the embodiments of the present application are clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments that can be derived from the embodiments given herein by a person of ordinary skill in the art are intended to be within the scope of the present disclosure.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure as claimed. Unless defined otherwise, all technical and scientific terms used herein have the same technical meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise, and it should be further understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of the stated features, steps, operations, devices, components, and/or combinations thereof.
An airport passenger security inspection dangerous goods real-time high-identification-rate method based on X-ray machine images comprises the following steps:
s1: constructing a dangerous article shape characteristic sample library and constructing a dangerous article color model sample;
s2: classifying and labeling the image data of the shape sample of the dangerous goods, and dividing the classified and labeled data into a training set, a verification set and a test set;
s3: establishing a deep learning neural network model;
s4: training a neural network model, and determining network model parameters; inputting the training set serving as a sample into a neural network model for training, inputting the verification set serving as a sample into the neural network model, and verifying the accuracy of the neural network model; initial values of weight parameters and deviation parameters of the neural network model are given, and the training speed is improved; secondly, adjusting a network architecture, judging the effectiveness of training through the variation trend of the minimum variance sum of a training set and a verification set, if the loss of the training set continuously decreases, the loss of the verification set also continuously decreases, which indicates that the neural network continues to learn normally, and if the loss of the training set continuously decreases, the loss of the verification set tends to be unchanged, which indicates that the network is overfitting, so that the training can be stopped, and determining the weight parameters and the deviation parameters of the neural network;
s5: testing the trained neural network through a test set to set a type discrimination detection threshold and a positioning threshold of the dangerous goods; the threshold parameter is adjusted to meet the requirement that the actual project is free from false detection and missing detection, and the detection rate of the system is guaranteed to the maximum extent while the missing detection is avoided.
S6: combining the dangerous goods color model with a deep learning neural network to judge the type of the dangerous goods; firstly, judging through a color model, and detecting by a deep learning method if no problem exists; the specific determination method of the color model is as follows: by converting the collected RGB image into an HSI image and combining threshold judgment of two color models, when the R channel numerical value and the brightness are greater than the set threshold and the image area is greater than the set threshold, an alarm is directly given, or when the image threshold of the RGB three channels is less than the set threshold, an alarm is given. The method is characterized in that dangerous goods are recognized according to learning that dangerous goods are mainly in shape features, and dangerous goods without obvious shape features cannot be recognized.
The dangerous article shape characteristic sample library is obtained by purchasing a dangerous article sample library, and a sample image library is constructed in a targeted manner, so that samples can be acquired from multiple visual angles. On one hand, the acquired sample image has complete and effective information, and on the other hand, the efficiency of labeling the sample image can be improved. Although the method is simple and quick, the method has the defects that the labeling is too extensive, background interference is introduced, and the classification and identification accuracy rate is reduced for the training and learning effect.
The dangerous goods color model separates the collected X-ray image into an RGB three-channel color model, converts the RGB three-channel color model into an HIS color model, and takes the dangerous goods color model component relation of the image as a criterion for identifying dangerous goods; the method specifically comprises the following steps: firstly, an acquired three-channel X-ray image is separated into R, G, B three-channel color model images, then the three-channel color model images are converted into HIS color model images according to the images of the RGB three-channel color model, a saturation threshold value in the image HIS model is assumed to be ST, a red component threshold value in the RGB color model is assumed to be RT, and the images meeting the conditions of R > G > B and S > = (255-R) × ST/RT and continuous areas of more than 50 pixels are combined to serve as criteria of a dangerous goods color model.
And the dangerous goods color sample is collected by adopting an X-ray machine to collect images to establish an image sample library, and the image sample library contains a certain amount of images of the target object.
The classification in S2 is based on whether there is an obvious contour distinguishing feature, and if the contour can be distinguished from the outline, the entire contour is used for distinguishing; if the outline can not be distinguished, the outline of the part is used for distinguishing; and the classification labeling in the S2 adopts a classification labeling method of multi-key point characteristics. For the image generated by the X-ray machine is different from a color image or a gray image generated by a general camera on the surface of an object, the X-ray image is only a pseudo color image based on different degrees of absorption of X-rays by internal tissues of the object, and the original data is an image with different gray levels. Firstly, images formed by different internal tissues of dangerous goods are distinguished, and bases of object classification features are sequentially made: the cutter is basically metal and has obvious profile characteristics, and the cutter adopts the overall profile characteristics and the color characteristics as the reference of depth learning and marking; the lighter is various, especially X-ray image is difficult to distinguish the discernment to it from the appearance profile, divide into electronic lighter and flint lighter to classify and mark for the convenience of discernment training with the lighter, though the flint lighter can divide into flint gas lighter and flint kerosene lighter again, its common characteristic still is flint and striking light characteristic part, so the mark benchmark that adopts the electronic lighter is electron striking light part profile characteristic, the mark benchmark that the flint lighter adopted is flint striking light part profile characteristic.
In S4, if the training of the network is normal and the loss of the training set and the verification set is less than 5, the network learning may be stopped, and if the loss of the training set and the loss of the verification set are both less than the set threshold and the number of times of non-convergence is greater than the set threshold, the training of the network model may also be stopped.
The deep learning neural network model in the S3 is a neural network model which is more suitable for X-ray image detection, and firstly, the size of an X-ray image is subjected to lossless labeling, data enhancement and training; secondly, a Darnet53 neural network is used as a main training network, in order to prevent false detection rate, as the object distance of an X-ray image is basically fixed and basically does not zoom, dangerous goods can be quickly detected by reducing multi-scale training parameters, meanwhile, the detection accuracy of the dangerous goods is ensured, and finally, the position of a target is positioned and the dangerous goods are classified.
In order to meet the increase of dangerous article types, new samples are inevitably generated in the actual process, the sample capacity of the samples which do not exist and dangerous articles which cannot be accurately identified needs to be enlarged, and the enlarged samples have good expansibility and easy maintainability in the application process of the program through a training model in an online incremental mode.
The invention discloses a multi-feature fusion data classification processing method based on image contour, color information and local feature combination. According to the combination of the contours, the color information and the local features of different dangerous goods images, the training is more targeted and flexible, and the detection accuracy of the system is effectively improved.
And labeling the sample image by adopting a polygonal feature surrounding method capable of defining multiple key points. The main characteristics of the sample can be extracted, background interference caused by the fact that a rectangular frame mark is adopted conventionally is filtered out, and accurate detection of dangerous goods is facilitated.
The detection rate of dangerous goods is improved by adopting a method of 'minor class training and major class identification'. The subclasses related to a large class of dangerous goods have respective characteristics, and the characteristics of the subclasses can be acquired more finely through subclass training, so that the training speed and the recognition accuracy are increased.
And the alarm processing is carried out by adopting a low alarm threshold value, so that the missing report rate of the system is effectively improved. At present, the identification and detection of dangerous goods based on images can not realize complete automation, and the method is also an auxiliary means, the principle of the auxiliary means is also 'Ning error detection and non-missing detection', and the method requires to reduce the alarm threshold, prevent the false alarm of the dangerous goods and improve the reliability of the detection of the dangerous goods.
By adopting the online incremental model learning and training method, the existing knowledge is well inherited, and the online incremental model learning and training method has good maintainability and expansibility. The auxiliary dangerous article detection system needs to expand dangerous article samples to detect confirmed dangerous articles so as to adapt to increase of dangerous article types, needs a large amount of time for retraining a large amount of training samples, saves training time for improving training efficiency, adopts an online incremental training method, can meet the requirement of online application training and learning, can inherit existing training parameters on one hand, can perform good optimization on the parameters, and can effectively improve the correct detection rate of the dangerous articles.
The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the present invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive efforts by those skilled in the art based on the technical solution of the present invention.
Claims (7)
1. An airport passenger security inspection dangerous goods real-time high-recognition rate method based on X-ray machine images is characterized in that: the method comprises the following steps:
s1: constructing a dangerous article shape characteristic sample library and constructing a dangerous article color model sample;
s2: classifying and labeling the image data of the shape sample of the dangerous goods, and dividing the classified and labeled data into a training set, a verification set and a test set;
s3: establishing a deep learning neural network model;
s4: training a neural network model, and determining network model parameters; inputting the training set serving as a sample into a neural network model for training, inputting the verification set serving as a sample into the neural network model, and verifying the accuracy of the neural network model; initial values of weight parameters and deviation parameters of the neural network model are given, and the training speed is improved; secondly, adjusting a network architecture, judging the effectiveness of training through the variation trend of the minimum variance sum of a training set and a verification set, if the loss of the training set continuously decreases, the loss of the verification set also continuously decreases, which indicates that the neural network continues to learn normally, and if the loss of the training set continuously decreases, the loss of the verification set tends to be unchanged, which indicates that the network is overfitting, so that the training can be stopped, and determining the weight parameters and the deviation parameters of the neural network;
s5: testing the trained neural network through a test set to set a type discrimination detection threshold and a positioning threshold of the dangerous goods;
s6: and (3) combining the dangerous article color model with a deep learning neural network to judge the type of the dangerous article: firstly, judging through a color model, and detecting by a deep learning method if no problem exists; the specific determination method of the color model is as follows: by converting the collected RGB image into an HSI image and combining threshold judgment of two color models, when the R channel numerical value and the brightness are greater than the set threshold and the image area is greater than the set threshold, an alarm is directly given, or when the image threshold of the RGB three channels is less than the set threshold, an alarm is given.
2. The airport passenger security check hazardous material real-time high-identification rate method based on X-ray machine image as claimed in claim 1, characterized in that: the dangerous article shape characteristic sample library is obtained by purchasing a dangerous article sample library, and a sample image library is constructed in a targeted manner, so that samples can be acquired from multiple visual angles.
3. The airport passenger security check hazardous material real-time high-identification rate method based on X-ray machine image as claimed in claim 1, characterized in that: the dangerous goods color model separates the collected X-ray image into an RGB three-channel color model, converts the RGB three-channel color model into an HIS color model, and takes the dangerous goods color model component relation of the image as a criterion for identifying dangerous goods; the method specifically comprises the following steps: firstly, an acquired three-channel X-ray image is separated into R, G, B three-channel color model images, then the three-channel color model images are converted into HIS color model images according to the images of the RGB three-channel color model, a saturation threshold value in the image HIS model is assumed to be ST, a red component threshold value in the RGB color model is assumed to be RT, and the images meeting the conditions of R > G > B and S > = (255-R) × ST/RT and continuous areas of more than 50 pixels are combined to serve as criteria of a dangerous goods color model.
4. The airport passenger security check hazardous material real-time high-identification rate method based on X-ray machine image as claimed in claim 1, characterized in that: and the dangerous goods color sample is collected by adopting an X-ray machine to collect images to establish an image sample library, and the image sample library contains a certain amount of images of the target object.
5. The airport passenger security check hazardous material real-time high-identification rate method based on X-ray machine image as claimed in claim 1, characterized in that: the classification in S2 is based on whether there is an obvious contour distinguishing feature, and if the contour can be distinguished from the outline, the entire contour is used for distinguishing; if the outline can not be distinguished, the outline of the part is used for distinguishing; and the classification labeling in the S2 adopts a classification labeling method of multi-key point characteristics.
6. The airport passenger security check hazardous material real-time high-identification rate method based on X-ray machine image as claimed in claim 1, characterized in that: in S4, if the training of the network is normal and the loss of the training set and the verification set is less than 5, the network learning may be stopped, and if the loss of the training set and the loss of the verification set are both less than the set threshold and the number of times of non-convergence is greater than the set threshold, the training of the network model may also be stopped.
7. The airport passenger security check hazardous material real-time high-identification rate method based on X-ray machine image as claimed in claim 1, characterized in that: the deep learning neural network model in S3 adopts a Darnet53 neural network as a backbone training network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010480894.4A CN111680729A (en) | 2020-05-30 | 2020-05-30 | Airport passenger security inspection dangerous goods real-time high-recognition rate method based on X-ray machine image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010480894.4A CN111680729A (en) | 2020-05-30 | 2020-05-30 | Airport passenger security inspection dangerous goods real-time high-recognition rate method based on X-ray machine image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111680729A true CN111680729A (en) | 2020-09-18 |
Family
ID=72452880
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010480894.4A Pending CN111680729A (en) | 2020-05-30 | 2020-05-30 | Airport passenger security inspection dangerous goods real-time high-recognition rate method based on X-ray machine image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111680729A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112444889A (en) * | 2020-11-13 | 2021-03-05 | 北京航星机器制造有限公司 | Rapid security inspection luggage remote centralized interpretation system and method |
| CN114693548A (en) * | 2022-03-08 | 2022-07-01 | 电子科技大学 | Dark channel defogging method based on bright area detection |
| CN114758259A (en) * | 2022-06-15 | 2022-07-15 | 科大天工智能装备技术（天津）有限公司 | Package detection method and system based on X-ray object image recognition |
| CN114758363A (en) * | 2022-06-16 | 2022-07-15 | 四川金信石信息技术有限公司 | Insulating glove wearing detection method and system based on deep learning |
| CN114814973A (en) * | 2022-03-28 | 2022-07-29 | 北京中盾安民分析技术有限公司 | Intelligent security check system and method for man-machine hybrid decision |
| CN116401587A (en) * | 2023-06-08 | 2023-07-07 | 乐山师范学院 | Object category identification method based on X-rays |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106886054A (en) * | 2017-04-13 | 2017-06-23 | 西安邮电大学 | Dangerous material automatic identification equipment and method based on 3 D X-ray imaging |
| CN207232762U (en) * | 2017-07-28 | 2018-04-13 | 广州大学华软软件学院 | A kind of disaster scene sniffing robot system |
| CN108182454A (en) * | 2018-01-18 | 2018-06-19 | 苏州大学 | Safety check identifying system and its control method |
| CN108198227A (en) * | 2018-03-16 | 2018-06-22 | 济南飞象信息科技有限公司 | Contraband intelligent identification Method based on X-ray screening machine image |
| CN111126447A (en) * | 2019-11-29 | 2020-05-08 | 中国船舶重工集团公司第七一三研究所 | Intelligent passenger security check luggage image automatic identification method |
-
2020
- 2020-05-30 CN CN202010480894.4A patent/CN111680729A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106886054A (en) * | 2017-04-13 | 2017-06-23 | 西安邮电大学 | Dangerous material automatic identification equipment and method based on 3 D X-ray imaging |
| CN207232762U (en) * | 2017-07-28 | 2018-04-13 | 广州大学华软软件学院 | A kind of disaster scene sniffing robot system |
| CN108182454A (en) * | 2018-01-18 | 2018-06-19 | 苏州大学 | Safety check identifying system and its control method |
| CN108198227A (en) * | 2018-03-16 | 2018-06-22 | 济南飞象信息科技有限公司 | Contraband intelligent identification Method based on X-ray screening machine image |
| CN111126447A (en) * | 2019-11-29 | 2020-05-08 | 中国船舶重工集团公司第七一三研究所 | Intelligent passenger security check luggage image automatic identification method |
Non-Patent Citations (1)
| Title |
|---|
| 武汉铁路局客运处: "《武汉铁路局客运规章文电汇编 2005-2013 上》", 31 December 2014 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112444889A (en) * | 2020-11-13 | 2021-03-05 | 北京航星机器制造有限公司 | Rapid security inspection luggage remote centralized interpretation system and method |
| CN114693548A (en) * | 2022-03-08 | 2022-07-01 | 电子科技大学 | Dark channel defogging method based on bright area detection |
| CN114693548B (en) * | 2022-03-08 | 2023-04-18 | 电子科技大学 | Dark channel defogging method based on bright area detection |
| CN114814973A (en) * | 2022-03-28 | 2022-07-29 | 北京中盾安民分析技术有限公司 | Intelligent security check system and method for man-machine hybrid decision |
| CN114814973B (en) * | 2022-03-28 | 2024-03-08 | 北京中盾安民分析技术有限公司 | Intelligent security inspection system and method for man-machine hybrid decision |
| CN114758259A (en) * | 2022-06-15 | 2022-07-15 | 科大天工智能装备技术（天津）有限公司 | Package detection method and system based on X-ray object image recognition |
| CN114758363A (en) * | 2022-06-16 | 2022-07-15 | 四川金信石信息技术有限公司 | Insulating glove wearing detection method and system based on deep learning |
| CN114758363B (en) * | 2022-06-16 | 2022-08-19 | 四川金信石信息技术有限公司 | Insulating glove wearing detection method and system based on deep learning |
| CN116401587A (en) * | 2023-06-08 | 2023-07-07 | 乐山师范学院 | Object category identification method based on X-rays |
| CN116401587B (en) * | 2023-06-08 | 2023-08-18 | 乐山师范学院 | Object category identification method based on X-rays |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111680729A (en) | Airport passenger security inspection dangerous goods real-time high-recognition rate method based on X-ray machine image | |
| CN110018524B (en) | X-ray security inspection contraband identification method based on vision-attribute | |
| CN113269073B (en) | Ship multi-target tracking method based on YOLO V5 algorithm | |
| CN108765412B (en) | Strip steel surface defect classification method | |
| CN104063722B (en) | A kind of detection of fusion HOG human body targets and the safety cap recognition methods of SVM classifier | |
| CN106446926A (en) | Transformer station worker helmet wear detection method based on video analysis | |
| CN110969130A (en) | Driver dangerous action identification method and system based on YOLOV3 | |
| CN106778833A (en) | Small object loses the automatic identifying method of failure under a kind of complex background | |
| CN108564069A (en) | A kind of industry safe wearing cap video detecting method | |
| US8068132B2 (en) | Method for identifying Guignardia citricarpa | |
| CN113553977B (en) | Improved YOLO V5-based safety helmet detection method and system | |
| CN113642474A (en) | Hazardous area personnel monitoring method based on YOLOV5 | |
| CN112906481A (en) | Method for realizing forest fire detection based on unmanned aerial vehicle | |
| CN107992854A (en) | Forest Ecology man-machine interaction method based on machine vision | |
| CN114782897A (en) | Dangerous behavior detection method and system based on machine vision and deep learning | |
| CN114662208B (en) | Construction visualization system and method based on Bim technology | |
| CN110458093A (en) | A kind of Safe belt detection method and corresponding equipment based on driver's monitoring system | |
| CN114841920A (en) | Flame identification method and device based on image processing and electronic equipment | |
| CN107330441A (en) | Flame image foreground extraction algorithm | |
| CN105469099B (en) | Pavement crack detection and identification method based on sparse representation classification | |
| CN113221667B (en) | Deep learning-based face mask attribute classification method and system | |
| CN105448095B (en) | Method and apparatus are surveyed in a kind of yellow mark car test | |
| CN103258218A (en) | Matte detection frame generation method and device and defect detection method and device | |
| CN112734702A (en) | Bridge safety monitoring method based on big data analysis and machine vision and cloud platform | |
| KR20210122429A (en) | Method and System for Artificial Intelligence based Quality Inspection in Manufacturing Process using Machine Vision Deep Learning |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200918 |
|
| RJ01 | Rejection of invention patent application after publication |