CN108960320A - Train coupling cock fault picture real-time detection method - Google Patents
Train coupling cock fault picture real-time detection method Download PDFInfo
- Publication number
- CN108960320A CN108960320A CN201810706143.2A CN201810706143A CN108960320A CN 108960320 A CN108960320 A CN 108960320A CN 201810706143 A CN201810706143 A CN 201810706143A CN 108960320 A CN108960320 A CN 108960320A
- Authority
- CN
- China
- Prior art keywords
- image
- feature
- cock
- train
- detection method
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Train Traffic Observation, Control, And Security (AREA)
Abstract
The present invention relates to image procossings and fault identification field, disclose a kind of train coupling cock fault picture real-time detection method, include the following steps: using image making database, extract multiple dimensioned union feature, it is normalized, and then training cascade detectors, then angle cock image to be detected is acquired, feature is sent into trained cascade detectors after extracting multiple dimensioned union feature and being normalized, positioning target area simultaneously calculates confidence level, compared with the threshold value of setting, detect whether angle cock breaks down.Train coupling cock fault picture real-time detection method of the present invention, detection efficiency and accuracy rate are high, and strong real-time, can find angle cock failure in time.
Description
Technical field
The present invention relates to image procossings and fault identification field, and in particular to a kind of train coupling cock fault picture is real-time
Detection method.
Background technique
In recent years, the technical inspection of train (referred to as " column inspection ") in China relied primarily on the manual detection mode of column inspection person and is
It is main.Which is influenced by artificial subjective factor such as eyesight, degree of fatigue and external environment such as weather, illumination etc., so as to cause inspection
It surveys efficiency and accuracy rate is influenced to different extents.Nowadays the high-speed rail and electric business in China flourish, logistics and transportation industry
Earth-shaking variation has occurred, railway transportation is in national economy as logistics and transportation industry important component
The status of mainstay, higher requirements are also raised for the following railway traffic safety problem.Then, TFDS (Trouble
Of moving Freight car Detection System, train operation fault dynamic images detection system) gradually apply
In on Some Domestic trunk railway.The system appliance computer, network communication, automatic control and image acquisition and processing technology are simultaneously drawn
Into the management method of science and the development approach of systematization, failure pictorial information dynamic is provided for the detection of railroad train operation troubles
It collects, store, transmission and Warning Service, the level that raising column inspection operation quality and efficiency and vehicle safety are taken precautions against are reinforced arranging
The man-machine system that failure basic information is collected, managed in vehicle utilization.The system can capture all figures of vehicle bottom and lower part automatically
Piece mainly delays the devices such as accessory, chassis, bogie, car body side lower part, angle cock comprising train hook, by man-computer cooperation side
Formula analyzes the content that can be photographed, and can differentiate that the device of vehicle has the failures such as N/D, fracture, loss, thus real
The leap that artificial detection is detected to man-computer cooperation is showed.
Angle cock is a key components and parts of Train Air Brake System, and train passes through the angle cock on main pipeline
Compressed air is transmitted to each section compartment, and is braked using compressed air.Only angle cock is opened, compressed air
It can be just transmitted to each section compartment, guarantee that train supervisor's is unimpeded.If train in the process of moving close by angle cock, this meeting
It causes a serious accident.Although domestic many scholars have carried out many relevant researchs, angle cock to the detection of TFDS failure
Region is too small for opposite train, it is difficult to position, in addition the complexity of angle cock background also gives the fault detection of angle cock
Increase a degree of difficulty.In recent years, related scholar proposes some different algorithms, as University Of Suzhou cares for superfine proposition
For detecting and identifying train bogie failure, southwestern traffic is big for a kind of shape representation and matching algorithm based on objective contour
Qin Na et al. is learned for the feature extraction of train bogie fault-signal, the estimation of critical component performance degradation and multiple features fusion and drop
The problems such as dimension, proposes that the feature extraction of bogie fault-signal and analytical framework, data-driven version are examined to solve bogie failure
The estimation of disconnected and component forms provides a kind of Research Thinking, but these algorithms are complicated, and real-time is not strong, used in angle cock
It is not very practical in fault detection.
Summary of the invention
The purpose of the present invention is to the deficiency of above-mentioned technology, a kind of train coupling cock fault picture is provided and is examined in real time
Survey method, detection efficiency and accuracy rate are high, and strong real-time, can find angle cock failure in time.
To achieve the above object, the train coupling cock fault picture real-time detection method designed by the present invention, including such as
Lower step:
A) railroad train operation troubles Motion Image Detection system original image collected is pre-processed, eliminates and claps
Influence of the environment to original image is taken the photograph, and the image after denoising is labeled, is made as the image data base with mark；
B) to the step A) production image data base extract characteristics of image, use swift nature pyramid extract image
Multiple dimensioned union feature, and carry out feature normalization processing；
C the step B) is utilized) grade of feature normalization treated the image data base training based on SVM-Adaboost
Join detector；
D image to be detected that angle cock) is acquired by railroad train operation troubles Motion Image Detection system, uses cunning
Dynamic window is slided on each layer of image to be detected pyramid with different step-lengths, and generates a series of wickets to be detected, using fast
Fast feature pyramid extracts the multiple dimensioned union feature of each wicket to be detected, and is normalized；
E the feature after normalized in the step D)) is sent into the step C) in trained cascade detectors,
Positioning target area simultaneously calculates confidence level；
F the threshold value comparison of calculated confidence level and setting in the step E)) is retained into the step if more than threshold value
Rapid E) in position target-region locating frame, angle cock is without failure, otherwise deletes the target-region locating frame, dog-ear
Cock breaks down.
Preferably, the step A) in, original image is to be acquired by train fault rail edge image detection system outdoor images
Equipment captured in real-time is transmitted by way of train bottom actuators dynamic image, and by fiber optic network.
Preferably, the step A) in, the image target area of positive sample is labeled, then positive negative sample is made respectively
It is made the angle cock image data base with mark, positive sample is non-faulting image, and negative sample is fault picture, angle cock
The production of image data base is to be stored separately fault picture and non-faulting image.
Preferably, the step B) in, multiple dimensioned union feature include 1 gradient magnitude, 6 histograms of oriented gradients and
Image Multiscale union feature is extracted using swift nature pyramid, and carries out feature and returns in totally 8 channels 1 invariable rotary LBP
One includes: the step of changing processing
A) union feature of a scale in every eight scales is accurately calculated using swift nature pyramid；
B) feature of other scale images eight scales Nei is calculated using this feature；
C) union feature extracted to eight scales is normalized.
Preferably, the step C) in, the cascade detectors are formed by several strong classifier set, the strong classification
Device is made of one group of SVM Weak Classifier, and when training cascade detectors, interior recycling AdaBoost iterative algorithm trains each
Strong classifier, outer circulation are the cascade detectors that training is made of all strong classifiers.
Preferably, the step F) in, confidence threshold value is 0.1~0.4.
Compared with prior art, the present invention having the advantage that by establishing image data base, it is special to extract multiple dimensioned joint
Levy and normalized, thus training cascade detectors, then by cascade detectors to the normalized of image to be detected after
Multiple dimensioned union feature detected, can detect in time whether angle cock breaks down, detection efficiency and accuracy rate
It is high.
Detailed description of the invention
Fig. 1 is the flow diagram of train coupling cock fault picture real-time detection method of the present invention；
Fig. 2 is the schematic diagram of angle cock non-faulting image in the present invention；
Fig. 3 is the schematic diagram of angle cock fault picture in the present invention；
Fig. 4 is that the multiple dimensioned union feature of angle cock in the present invention extracts flow diagram；
Fig. 5 is the testing process schematic diagram based on SVM-Adaboost cascade detectors in the present invention.
Specific embodiment
The following further describes the present invention in detail with reference to the accompanying drawings and specific embodiments.
A kind of train coupling cock fault picture real-time detection method, as shown in Figure 1, method includes the following steps:
A) railroad train operation troubles Motion Image Detection system original image collected is pre-processed, eliminates and claps
Influence of the environment to original image is taken the photograph, original image is that equipment reality is acquired by train fault rail edge image detection system outdoor images
When shooting by way of train bottom actuators dynamic image, and by fiber optic network transmission, wherein to the image object of positive sample
Region is labeled, then positive negative sample is fabricated to the angle cock image data base with mark respectively, and positive sample is non-event
Hinder image, as shown in Fig. 2, negative sample is fault picture, as shown in figure 3, the production of angle cock image data base is by fault graph
Picture and non-faulting image are stored separately；
B characteristics of image) is extracted to the image data base of step A) production, as shown in figure 4, mentioning using swift nature pyramid
The multiple dimensioned union feature of image is taken, and carries out feature normalization processing, in the present embodiment, multiple dimensioned union feature includes 1
Totally 8 channels gradient magnitude, 6 histograms of oriented gradients and 1 invariable rotary LBP use swift nature pyramid to extract figure
As multiple dimensioned union feature, and the step of carrying out feature normalization processing includes:
A) union feature (reference characteristic) of a scale in every eight scales is accurately calculated using swift nature pyramid；
B) feature of other scale images eight scales Nei is calculated using this feature；
C) union feature extracted to eight scales is normalized.
Specifically, since the image of collected angle cock is grayscale image, therefore including to image selection in the present embodiment
The feature in totally 8 channels 1 gradient magnitude, 6 histograms of oriented gradients and 1 invariable rotary LBP, swift nature pyramid exist
Under the conditions of not losing key message, the extraction rate of feature can be greatly speeded up.The pyramidal realization process of swift nature is as follows:
IsIt indicates that image of the image I at scale s, R (I, s) indicate that image I carries out s resampling, defines one with flat
The function Ω of motion immovability extracts the channel characteristics C of s resamplings=Ω (R (I, s)), and use CsSuccessively calculated, in order to
Accelerate speed and simplify to calculate, to CsIt can carry out following approximate calculation:
In formula, C=Ω (I) is the channel characteristics of image I, λΩFor a characteristic parameter, λΩIt can be according to specific channel
Feature distribution is fitted to obtain.
One feature pyramid is the multi-scale Representation to image I, then extracts the corresponding characteristics of image of each scale.Ruler
S is spent since 1, is equidistantly sampled in the space log, is that 4~12 are chosen in eight scale (octave) under normal circumstances
The interval of scale, each eight scale is the half at previous eight scale interval.Swift nature pyramid is established, is first calculated every
The feature C of a eight scales images′=Ω (R (I, s ')),The spy of image between eight scales is calculated with formula again
Sign, formulaBecomeWherein s is the nearest ruler of distance s '
Degree,Swift nature pyramid only calculates the feature of a scale in each eight scale,
Reuse the feature that this feature calculates other scale images eight scale Nei again.
After 8 scale features for extracting image, the feature dimension of 8 scales is inconsistent, may result in detector below
Training can not converge to globally optimal solution, therefore, the feature of each scale is normalized, calculation is as follows:
Wherein, ‖ ‖2Indicate 2- norm；ε is the constant of a very little.
C step B) is utilized) grade joint inspection of feature normalization treated the image data base training based on SVM-Adaboost
Device is surveyed, cascade detectors are formed by several strong classifier set, and strong classifier is made of one group of SVM Weak Classifier, training grade
When joining detector, each strong classifier of interior recycling AdaBoost iterative algorithm training, outer circulation is training by all strong
The cascade detectors of classifier composition, in the present embodiment, training process is as follows:
Step 1: the whole false positive rate F of initialized targettarget, the false positive rate f of the maximum of strong classifier stagemaxIt is examined with minimum
Survey rate dmin, positive sample P and negative sample N；
Step 2: interior circulation: training a strong classifier stage using AdaBoost algorithm, often adds in this circulation
The threshold value that next SVM Weak Classifier is added is readjusted while entering a SVM Weak Classifier, guarantees strong classifier stage
Verification and measurement ratio be not less than dmin, and assessment is re-started to the strong classifier, judge the positive rate f of the vacation of the strong classifieriWhether it is lower than
fmax, if it is, the strong classifier training of this level-one terminates, otherwise, continue to add SVM Weak Classifier, until meeting internal follow
The termination condition of ring；
Step 3: outer circulation: judge the whole false positive rate F of current goaliWhether lower than the positive rate of the whole vacation of initialized target
Ftarget, if it is, training terminates, otherwise, readjust negative sample and the sample of all detection mistakes be put into database N
In, then internal circulation is carried out, until the whole false positive rate F of current goaliFalse positive rate F whole lower than initialized targettargetUntil；
D image to be detected that angle cock) is acquired by railroad train operation troubles Motion Image Detection system, uses cunning
Dynamic window is slided on each layer of image to be detected pyramid with different step-lengths, and generates a series of wickets to be detected, with reference to step
Rapid B), the multiple dimensioned union feature of each wicket to be detected is extracted using swift nature pyramid, and be normalized；
E) by trained cascade detectors in the feature feeding step C after normalized in step D)), such as Fig. 5 institute
Show, position target area and calculates confidence level；
F) by the threshold value comparison of confidence level and setting calculated in step E), confidence threshold value is 0.1~0.4, this reality
Apply in example, threshold value takes 0.25, if more than threshold value, retains step E) in the target-region locating frame that positions, angle cock do not occur
Failure, otherwise delete target zone location frame, angle cock break down.
Claims (6)
1. a kind of train coupling cock fault picture real-time detection method, it is characterised in that: described method includes following steps:
A) railroad train operation troubles Motion Image Detection system original image collected is pre-processed, eliminates shooting ring
Influence of the border to original image, and the image after denoising is labeled, it is made as the image data base with mark；
B) to the step A) production image data base extract characteristics of image, use swift nature pyramid extract image it is more
Scale union feature, and carry out feature normalization processing；
C the step B) is utilized) grade joint inspection of feature normalization treated the image data base training based on SVM-Adaboost
Survey device；
D image to be detected that angle cock) is acquired by railroad train operation troubles Motion Image Detection system, uses sliding window
Mouth is slided on each layer of image to be detected pyramid with different step-lengths, and generates a series of wickets to be detected, uses quick spy
Sign pyramid extracts the multiple dimensioned union feature of each wicket to be detected, and is normalized；
E) by trained cascade detectors, positioning in the feature feeding step C after normalized in the step D))
Target area simultaneously calculates confidence level；
F the threshold value comparison of calculated confidence level and setting in the step E)) is retained into the step E if more than threshold value)
The target-region locating frame of middle positioning, angle cock is without failure, otherwise deletes the target-region locating frame, angle cock
It breaks down.
2. train coupling cock fault picture real-time detection method according to claim 1, it is characterised in that: the step A)
In, original image is that equipment captured in real-time is acquired by train fault rail edge image detection system outdoor images by way of train bottom system
Dynamic device dynamic image, and transmitted by fiber optic network.
3. train coupling cock fault picture real-time detection method according to claim 1, it is characterised in that: the step A)
In, the image target area of positive sample is labeled, then positive negative sample is fabricated to the angle cock figure with mark respectively
As database, positive sample is non-faulting image, and negative sample is fault picture, and the production of angle cock image data base is by fault graph
Picture and non-faulting image are stored separately.
4. train coupling cock fault picture real-time detection method according to claim 1, it is characterised in that: the step B)
In, multiple dimensioned union feature includes totally 8 channels 1 gradient magnitude, 6 histograms of oriented gradients and 1 invariable rotary LBP,
Image Multiscale union feature is extracted using swift nature pyramid, and the step of carrying out feature normalization processing includes:
A) union feature of a scale in every eight scales is accurately calculated using swift nature pyramid；
B) feature of other scale images eight scales Nei is calculated using this feature；
C) union feature extracted to eight scales is normalized.
5. train coupling cock fault picture real-time detection method according to claim 1, it is characterised in that: the step C)
In, the cascade detectors are formed by several strong classifier set, and the strong classifier is made of one group of SVM Weak Classifier,
When training cascade detectors, each strong classifier of interior recyclings AdaBoost iterative algorithm training, outer circulation be it is trained by
The cascade detectors of all strong classifier compositions.
6. train coupling cock fault picture real-time detection method according to claim 1, it is characterised in that: the step F)
In, confidence threshold value is 0.1~0.4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810706143.2A CN108960320B (en) | 2018-06-26 | 2018-06-26 | Real-time detection method for fault image of angle cock of train |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810706143.2A CN108960320B (en) | 2018-06-26 | 2018-06-26 | Real-time detection method for fault image of angle cock of train |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108960320A true CN108960320A (en) | 2018-12-07 |
| CN108960320B CN108960320B (en) | 2021-04-27 |
Family
ID=64484846
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810706143.2A Active CN108960320B (en) | 2018-06-26 | 2018-06-26 | Real-time detection method for fault image of angle cock of train |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108960320B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112733706A (en) * | 2021-01-07 | 2021-04-30 | 河南理工大学 | Motor fault diagnosis method based on bilinear LBP, electronic equipment and medium |
| CN113830136A (en) * | 2021-10-20 | 2021-12-24 | 哈尔滨市科佳通用机电股份有限公司 | Method for identifying malposition fault of angle cock handle of railway wagon |
| CN115170828A (en) * | 2022-07-15 | 2022-10-11 | 哈尔滨市科佳通用机电股份有限公司 | Angle cock clip loss fault detection method based on deep learning |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105260744A (en) * | 2015-10-08 | 2016-01-20 | 北京航空航天大学 | Automatic on-line diagnosis method for freight train coupler tail cotter position faults and system |
| CN105260749A (en) * | 2015-11-02 | 2016-01-20 | 中国电子科技集团公司第二十八研究所 | Real-time target detection method based on oriented gradient two-value mode and soft cascade SVM |
| CN105447511A (en) * | 2015-11-15 | 2016-03-30 | 安徽清新互联信息科技有限公司 | SVM object detection method based on Adaboost Haar-Like features |
| CN106485694A (en) * | 2016-09-11 | 2017-03-08 | 西南交通大学 | A kind of high ferro contact net double-jacket tube connector six-sided nut based on cascade classifier comes off defective mode detection method |
| CN106778833A (en) * | 2016-11-28 | 2017-05-31 | 北京航天控制仪器研究所 | Small object loses the automatic identifying method of failure under a kind of complex background |
| US20180046151A1 (en) * | 2015-03-11 | 2018-02-15 | Siemens Indsutry, Inc. | Cascaded identification in building automation |
-
2018
- 2018-06-26 CN CN201810706143.2A patent/CN108960320B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180046151A1 (en) * | 2015-03-11 | 2018-02-15 | Siemens Indsutry, Inc. | Cascaded identification in building automation |
| CN105260744A (en) * | 2015-10-08 | 2016-01-20 | 北京航空航天大学 | Automatic on-line diagnosis method for freight train coupler tail cotter position faults and system |
| CN105260749A (en) * | 2015-11-02 | 2016-01-20 | 中国电子科技集团公司第二十八研究所 | Real-time target detection method based on oriented gradient two-value mode and soft cascade SVM |
| CN105447511A (en) * | 2015-11-15 | 2016-03-30 | 安徽清新互联信息科技有限公司 | SVM object detection method based on Adaboost Haar-Like features |
| CN106485694A (en) * | 2016-09-11 | 2017-03-08 | 西南交通大学 | A kind of high ferro contact net double-jacket tube connector six-sided nut based on cascade classifier comes off defective mode detection method |
| CN106778833A (en) * | 2016-11-28 | 2017-05-31 | 北京航天控制仪器研究所 | Small object loses the automatic identifying method of failure under a kind of complex background |
Non-Patent Citations (2)
| Title |
|---|
| ZHOU FUQIANG, ET AL: "《Automated visual inspection of angle cocks during train operation》", 《PROCEEDINGS OF THE INSTITUTION OF MECHANICAL ENGINEERS, PART F: JOURNAL OF RAIL AND RAPID TRANSIT 》 * |
| 孙国栋 等31: "《基于多特征融合和BP-AdaBoost算法的列车关键零件故障自动识别》", 《中国机械工程》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112733706A (en) * | 2021-01-07 | 2021-04-30 | 河南理工大学 | Motor fault diagnosis method based on bilinear LBP, electronic equipment and medium |
| CN112733706B (en) * | 2021-01-07 | 2022-09-16 | 河南理工大学 | Motor fault diagnosis method based on bilinear LBP, electronic equipment and medium |
| CN113830136A (en) * | 2021-10-20 | 2021-12-24 | 哈尔滨市科佳通用机电股份有限公司 | Method for identifying malposition fault of angle cock handle of railway wagon |
| CN113830136B (en) * | 2021-10-20 | 2022-04-19 | 哈尔滨市科佳通用机电股份有限公司 | Method for identifying malposition fault of angle cock handle of railway wagon |
| CN115170828A (en) * | 2022-07-15 | 2022-10-11 | 哈尔滨市科佳通用机电股份有限公司 | Angle cock clip loss fault detection method based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108960320B (en) | 2021-04-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105260744B (en) | The automatic on-line diagnostic method and system of a kind of goods train coupler yoke key position failure | |
| CN102306290B (en) | Face tracking recognition technique based on video | |
| CN111126325B (en) | Intelligent personnel security identification statistical method based on video | |
| CN101038686B (en) | Method for recognizing machine-readable travel certificate | |
| CN103279756B (en) | Vehicle detection based on integrated classifier analyzes system and determination method thereof | |
| CN105424330B (en) | A kind of lorry bolts at axle end fault detection method and device | |
| CN108205661A (en) | A kind of ATM abnormal human face detection based on deep learning | |
| Hu et al. | Detection of workers without the helments in videos based on YOLO V3 | |
| CN106250936A (en) | Multiple features multithreading safety check contraband automatic identifying method based on machine learning | |
| CN104881865A (en) | Forest disease and pest monitoring and early warning method and system based on unmanned plane image analysis | |
| CN108960320A (en) | Train coupling cock fault picture real-time detection method | |
| CN108985170A (en) | Transmission line of electricity hanger recognition methods based on Three image difference and deep learning | |
| CN113947731B (en) | Foreign matter identification method and system based on contact net safety inspection | |
| CN103745224A (en) | Image-based railway contact net bird-nest abnormal condition detection method | |
| CN108062542A (en) | The detection method for the face being blocked | |
| CN106650623A (en) | Face detection-based method for verifying personnel and identity document for exit and entry | |
| CN106548131A (en) | A kind of workmen's safety helmet real-time detection method based on pedestrian detection | |
| CN112183472A (en) | Method for detecting whether test field personnel wear work clothes or not based on improved RetinaNet | |
| CN109993061A (en) | A kind of human face detection and tracing method, system and terminal device | |
| CN109829458A (en) | The method of the journal file of record system operatio behavior is automatically generated in real time | |
| CN110009058A (en) | A kind of parking lot Vehicle License Plate Recognition System and method | |
| CN111339927B (en) | Intelligent recognition system for working state of personnel in electric power business hall | |
| CN113673614B (en) | Metro tunnel foreign matter intrusion detection device and method based on machine vision | |
| CN107944373A (en) | A kind of video anomaly detection method based on deep learning | |
| CN105447463B (en) | Across the camera to automatically track system that substation is identified based on characteristics of human body |
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 |