CN109344886A - Occlusion number plate distinguishing method based on convolutional neural network - Google Patents
Occlusion number plate distinguishing method based on convolutional neural network Download PDFInfo
- Publication number
- CN109344886A CN109344886A CN201811081354.8A CN201811081354A CN109344886A CN 109344886 A CN109344886 A CN 109344886A CN 201811081354 A CN201811081354 A CN 201811081354A CN 109344886 A CN109344886 A CN 109344886A
- Authority
- CN
- China
- Prior art keywords
- number plate
- vehicle
- classification
- picture
- convolutional neural
- 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
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Probability & Statistics with Applications (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention provides a convolutional neural network-based shielding number plate distinguishing method, which can continuously improve the identification capability and ensure that higher identification accuracy can be maintained. It includes: s1, extracting data from the road monitoring equipment through an image recognition technology, and outputting a vehicle region picture; s2, detecting the number plate of the vehicle, and marking the vehicle without the detected number plate as a suspect vehicle; s3, obtaining a candidate area picture for judging the shielding license plate; s4, periodically and circularly collecting license plate samples; s5, extracting all license plate samples through an image recognition technology, and outputting vehicle region pictures of the samples; s6, obtaining a candidate area picture for judging the shielding license plate; s7, after a new license plate sample is collected each time, a convolutional neural network is trained; and S8, inputting the obtained candidate area picture of the suspected vehicle into a trained convolutional neural network, judging whether the license plate is shielded, and if the suspected vehicle is judged to have the behavior of shielding the license plate, entering a subsequent program aiming at the suspected vehicle.
Description
Technical field
The present invention relates to image identification technical fields in intellectual traffic control, specially blocking based on convolutional neural networks
Number plate method of discrimination.
Background technique
It blocks number plate and refers to the behavior for blocking number plate of vehicle using shelters such as cloth set, paper handkerchiefs, make existing monitoring
Equipment can not recognize its true number plate information.Number plate is blocked as vehicle and relates to one of board illegal activities, WeiZhao's Notes
It is lower, but make a very bad impression;Number plate is blocked due to the easy perceptibility of its naked eyes, is not a kind of long-term illegal activities, one it has been observed that
True number plate information can be investigated by Track association analysis, to scheme to search the technologies such as vehicle;But existing number plate recognition methods
Majority depends on number plate detection algorithm and Character segmentation algorithm, because in existing number plate detection algorithm, for detected material
Definition is most of be it is fixed, be not easy to change, due to the coming in every shape of shelter, not of uniform size and may change at any time,
When the definition in shelter and algorithm is inconsistent, frequent occurrence the problem of error in judgement.
Summary of the invention
In order to solve the problem of that car plate detection result is easy error when license plate shading state changes, the present invention is mentioned
It, can continual promotion recognition capability for blocking number plate method of discrimination based on convolutional neural networks, it is ensured that Neng Gouwei
Hold higher recognition accuracy.
The technical scheme is that such: blocking number plate method of discrimination based on convolutional neural networks comprising with
Lower step:
S1: by image recognition technology, extracted from road monitoring equipment monitoring picture perhaps video data and from picture or
Information of vehicles is obtained in video, detects vehicle region, exports vehicle region picture;
S2: carrying out number plate of vehicle detection in the vehicle region picture, is suspicion by the marking of cars for not detecting number plate
Vehicle;
It is characterized in that, its is further comprising the steps of:
S3: for the vehicle region picture of the suspected vehicles, the candidate region figure for blocking number plate differentiation is obtained
Piece;
S4: the circulating collection license plate sample of timing;
S5: extracting all license plate samples by image recognition technology, detects vehicle region, exports the license plate
The vehicle region picture of sample;
S6: for the vehicle region picture of all license plate samples, the candidate for blocking number plate differentiation is obtained
Region picture;
S7: after collecting the new license plate sample every time, the candidate region picture input convolution mind of the license plate sample
Through network, training convolutional neural networks;
S8: the candidate region picture of the suspected vehicles obtained in step S3, it is input to the institute obtained in step S7
It states in trained convolutional neural networks, determines whether to block number plate, if the row for blocking number plate is not present in the suspected vehicles
For then this differentiation terminates;If it is determined that there is the behavior for blocking number plate in the suspected vehicles, then for the suspected vehicles into
Enter down-stream.
It is further characterized by:
In step S3 and S6, the method for blocking the candidate region picture of number plate differentiation is obtained, is included the following steps:
A: the picture top half of the vehicle region is removed;
B: the lower half portion of the remaining vehicle region, the area of each removal 10% is originated from right boundary;
Remaining region is defined as candidate region, the picture in the candidate region is the candidate for blocking number plate differentiation
Region picture;
In the step s 7, the method for training convolutional neural networks includes the following steps:
S7a: a subseries is carried out to the candidate region picture of the license plate sample, the type of a subseries includes:
Partial occlusion number plate, it is unlicensed, other;
Partial occlusion number plate classification refers to that number slip of vehicle has been blocked the situation of a part;
The unlicensed classification does not hang number plate including vehicle and number slip of vehicle is blocked two kinds of situations completely;
Other described classification refer to that number slip of vehicle is in two kinds of partial occlusion number plate described above classification, unlicensed classification feelings
Other situations other than condition;
S7b: by three kinds of candidates to classify of partial occlusion number plate classification, the unlicensed classification, other classification
Region picture is input in convolutional neural networks, trains a 3 sorting algorithm models using convolutional neural networks, trained
Model is denoted as model X;
S7c: carrying out secondary classification for the candidate region picture for the license plate sample for including in the unlicensed classification, described secondary
The type that classification includes are as follows: do not hang number plate, block number plate completely;
S7d: number plate classification and the candidate region picture for blocking number plate classification completely of not hanging is input to volume
In product neural network, a 2 sorting algorithm models are trained using convolutional neural networks, trained model is denoted as model Y;
In step 8, the method for determining whether to block number slip, includes the following steps:
S8a: it the candidate region picture for the suspected vehicles for blocking number plate differentiation obtained in step S3, is input to
The trained model X carries out principium identification, and standard is as follows:
If the maximum probability and probability of partial occlusion are more than or equal to threshold value, assert in the presence of behavior is blocked, then assert to exist and block
Behavior;
If the maximum probability but probability of partial occlusion are less than threshold value, assert there is no behavior is blocked, exit judgement;
If other maximum probabilities, assert there is no number plate behavior is blocked, exit judgement;
If unlicensed maximum probability and probability is more than or equal to threshold value, secondary discrimination is carried out;
Unlicensed maximum probability and probability is less than threshold value and then assert there is no behavior is blocked, and exits judgement;
S8b: maximum probability and probability unlicensed in step S8a are more than or equal to the candidate of the suspected vehicles of threshold value
Region picture carries out the secondary discrimination, the method for the secondary discrimination as Target Photo are as follows:
The Target Photo is input in the model Y and is classified, classification standard is as follows:
If the maximum probability blocked completely, and probability is more than or equal to threshold value, then assert to exist and block behavior;
Otherwise assert that there is no block number plate behavior without exception.
It is provided by the invention to block number plate method of discrimination based on convolutional neural networks, by being based on convolutional neural networks
Image classification is carried out to the candidate region picture of suspected vehicles, finally judges that target vehicle whether there is and blocks behavior;Based on volume
The algorithm model generalization ability of product neural network is strong, also very good for the data fitting effect not in training set, using base
The identification for block in convolutional neural networks image classification method license plate behavior makes method of discrimination of the invention for each
When blocking behavior of seed type, it may have very high correctness;The positive negative sample of lasting collection license plate in application process, it is defeated
Enter into convolutional neural networks, it is lasting that convolutional neural networks are trained, it obtains changing synchronized update with actual conditions
Trained convolutional neural networks model further ensures that the identification of the method for discrimination in technical solution of the present invention is accurate
Rate.
Detailed description of the invention
Fig. 1 is the overall flow schematic diagram of this method;
Fig. 2 is the flow diagram for obtaining the candidate region picture for blocking number plate differentiation;
Fig. 3 is the flow diagram that secondary image classification is carried out by trained convolutional neural networks structure.
Specific embodiment
As shown in Fig. 1 ~ Fig. 3, the present invention is to block number plate method of discrimination based on convolutional neural networks comprising following step
It is rapid:
S1: by image recognition technology, from road monitoring equipment, as highway bayonet system in extract monitoring picture or
Video data, and information of vehicles is obtained from picture or video, vehicle region is detected, vehicle region picture is exported;
S2: carrying out number plate of vehicle detection in vehicle region picture, is suspected vehicles by the marking of cars for not detecting number plate;
S3: for the vehicle region picture of suspected vehicles, the candidate region picture for blocking number plate differentiation is obtained;
S4: what is regularly recycled collects a large amount of picture materials from various regions highway bayonet system, as license plate sample;By fixed
The collection license plate sample of the circulation of phase, it is lasting that convolutional neural networks are trained, obtain it is synchronous with actual conditions variation more
New trained convolutional neural networks further ensure that the identification of the method for discrimination in technical solution of the present invention is accurate
Rate;
S5: extracting all license plate samples by image recognition technology, detects vehicle region, exports the vehicle of license plate sample
Region picture;
S6: for the vehicle region picture of all license plate samples, the candidate region picture for blocking number plate differentiation is obtained;
S7: after collecting new affiliated license plate sheet every time, the candidate region picture of license plate sample is inputted convolutional neural networks, training
Convolutional neural networks;
S8: it the candidate region picture of suspected vehicles obtained in step S3, is input in trained convolutional neural networks, sentences
Fixed whether to block number plate, if the behavior for blocking number plate is not present in suspected vehicles, this differentiation terminates;If it is determined that suspicion vehicle
There is the behavior of number plate of blocking, then enters down-stream for suspected vehicles.
In step S3 and S6, the method for blocking the candidate region picture of number plate differentiation is obtained, is included the following steps:
A: the picture top half of vehicle region is removed;
B: the lower half portion of remaining vehicle region, the area of each removal 10% is originated from right boundary;
Remaining region is defined as candidate region, the picture in candidate region is the candidate region figure for blocking number plate differentiation
Piece;
By the definition to candidate region, uniformly delimit the target area for doing image recognition technology, no matter license plate shading object
Shape, size all only do image recognition within the scope of unified candidate region, and the part beyond candidate region is all not as target
Object;Avoid image recognition operation by license plate shading object shape, size interference caused by recognition result error it is biggish
The generation of problem;By the delimitation to candidate region, bring influence can not effectively be detected by reducing conventional number plate detection algorithm,
Operation difficulty is not only reduced, and improves the accuracy of identification.
In the step s 7, the method for training convolutional neural networks includes the following steps:
S7a: a subseries is carried out to the candidate region picture of license plate sample, the type of a subseries includes: partial occlusion number
Board, it is unlicensed, other;Wherein the unlicensed classification includes not hanging number plate, blocking two kinds of situations of number plate completely;
Partial occlusion number plate classification refer to number slip of vehicle be blocked a part situation;
Unlicensed classification does not hang number plate including vehicle and number slip of vehicle is blocked two kinds of situations completely;
Other classification refer to that number slip of vehicle is in other situations other than above-mentioned two situations;
S7b: partial occlusion number plate, unlicensed, other three kinds classification picture are input in convolutional neural networks, convolution mind is utilized
Go out a 3 sorting algorithm models through network training, trained model is denoted as model X;
S7c: the candidate region picture for the license plate sample for including in unlicensed classification is subjected to secondary classification, the type packet of secondary classification
It includes: not hanging number plate, blocks number plate completely;
S7d: number plate will not be hung, the pictures of two types of number plate is blocked completely and is input in convolutional neural networks, utilizes convolution
Neural metwork training goes out a 2 sorting algorithm models, and trained model is denoted as model Y.
In step 8, the method for determining whether to block number slip, includes the following steps:
S8a: it the candidate region picture for the suspected vehicles for blocking number plate differentiation obtained in step S3, is input to trained
Model X carries out principium identification, and standard is as follows:
If the maximum probability and probability of partial occlusion are more than or equal to threshold value, assert in the presence of behavior is blocked, then assert to exist and block
Behavior;
If the maximum probability but probability of partial occlusion are less than threshold value, assert there is no behavior is blocked, exit judgement;
If other maximum probabilities, assert there is no number plate behavior is blocked, exit judgement;
If unlicensed maximum probability and probability is more than or equal to threshold value, secondary discrimination is carried out;
Unlicensed maximum probability and probability is less than threshold value and then assert there is no behavior is blocked, and exits judgement;
S8b: maximum probability and probability unlicensed in step S8a are more than or equal to the candidate region picture of the suspected vehicles of threshold value
As Target Photo, secondary discrimination, the method for secondary discrimination are carried out are as follows: Target Photo is input in model Y and is classified, point
Class standard is as follows:
If the maximum probability blocked completely, and probability is more than or equal to threshold value, then assert to exist and block behavior;
Otherwise assert there is no number plate behavior is blocked, exit judgement;
By model X and model Y points to carry out discriminant classification to picture twice, more classification " the number of suspension of similar features
Board " and " blocking number plate entirely " are first merged into a classification, are distinguished with other two classification by preliminary classification, are then led to
Secondary classification is crossed to distinguish " not hanging number plate " and " blocking number plate entirely " two kinds of situations;Such secondary classification method, with
The classification method of " differentiation of four kinds of classification is completed in a sort operation " is compared, and has higher accuracy, further to protect
The accuracy of discriminating method of the invention is demonstrate,proved.
After technical solution of the present invention, by extracting monitoring data in existing road monitoring equipment, after progress
Continuous image analysis, without increasing new vision facilities, advantage of lower cost;Road is analyzed by road monitoring equipment initiative recognition
The suspicion travelled on face deliberately blocks number plate vehicle, greatly improves traffic management department and actively discovers to such illegal activities
Ability;License plate sample is regularly collected in the present invention, cyclically trains classification neural network model, makes image classification model
Classification standard is maintained closer to reality to the accuracy rate for blocking license plate behavior discrimination.
Claims (4)
1. blocking number plate method of discrimination based on convolutional neural networks comprising following steps:
S1: by image recognition technology, extracted from road monitoring equipment monitoring picture perhaps video data and from picture or
Information of vehicles is obtained in video, detects vehicle region, exports vehicle region picture;
S2: carrying out number plate of vehicle detection in the vehicle region picture, is suspicion by the marking of cars for not detecting number plate
Vehicle;
It is characterized in that, its is further comprising the steps of:
S3: for the vehicle region picture of the suspected vehicles, the candidate region figure for blocking number plate differentiation is obtained
Piece;
S4: the circulating collection license plate sample of timing;
S5: extracting all license plate samples by image recognition technology, detects vehicle region, exports the license plate
The vehicle region picture of sample;
S6: for the vehicle region picture of all license plate samples, the candidate for blocking number plate differentiation is obtained
Region picture;
S7: after collecting the new license plate sample every time, the candidate region picture input convolution mind of the license plate sample
Through network, training convolutional neural networks;
S8: the candidate region picture of the suspected vehicles obtained in step S3, it is input to the institute obtained in step S7
It states in trained convolutional neural networks, determines whether to block number plate, if the row for blocking number plate is not present in the suspected vehicles
For then this differentiation terminates;If it is determined that there is the behavior for blocking number plate in the suspected vehicles, then for the suspected vehicles into
Enter down-stream.
2. blocking number plate method of discrimination based on convolutional neural networks according to claim 1, it is characterised in that: in step S3
In S6, the method for blocking the candidate region picture of number plate differentiation is obtained, is included the following steps:
A: the picture top half of the vehicle region is removed;
B: the lower half portion of the remaining vehicle region, the area of each removal 10% is originated from right boundary;
Remaining region is defined as candidate region, the picture in the candidate region is the candidate for blocking number plate differentiation
Region picture.
3. blocking number plate method of discrimination based on convolutional neural networks according to claim 1, it is characterised in that: in step S7
In, the method for training convolutional neural networks includes the following steps:
S7a: a subseries is carried out to the candidate region picture of the license plate sample, the type of a subseries includes:
Partial occlusion number plate, it is unlicensed, other;
Partial occlusion number plate classification refers to that number slip of vehicle has been blocked the situation of a part;
The unlicensed classification does not hang number plate including vehicle and number slip of vehicle is blocked two kinds of situations completely;
Other described classification refer to that number slip of vehicle is in two kinds of partial occlusion number plate described above classification, unlicensed classification feelings
Other situations other than condition;
S7b: by three kinds of candidates to classify of partial occlusion number plate classification, the unlicensed classification, other classification
Region picture is input in convolutional neural networks, trains a 3 sorting algorithm models using convolutional neural networks, trained
Model is denoted as model X;
S7c: carrying out secondary classification for the candidate region picture for the license plate sample for including in the unlicensed classification, described secondary
The type that classification includes are as follows: do not hang number plate, block number plate completely;
S7d: number plate classification and the candidate region picture for blocking number plate classification completely of not hanging is input to volume
In product neural network, a 2 sorting algorithm models are trained using convolutional neural networks, trained model is denoted as model Y.
4. blocking number plate method of discrimination based on convolutional neural networks according to claim 1, it is characterised in that: in step 8
In, the method for determining whether to block number slip includes the following steps:
S8a: it the candidate region picture for the suspected vehicles for blocking number plate differentiation obtained in step S3, is input to
The trained model X carries out principium identification, and standard is as follows:
If the maximum probability and probability of partial occlusion are more than or equal to threshold value, assert in the presence of behavior is blocked, then assert to exist and block
Behavior;
If the maximum probability but probability of partial occlusion are less than threshold value, assert there is no behavior is blocked, exit judgement;
If other maximum probabilities, assert there is no number plate behavior is blocked, exit judgement;
If unlicensed maximum probability and probability is more than or equal to threshold value, secondary discrimination is carried out;
Unlicensed maximum probability and probability is less than threshold value and then assert there is no behavior is blocked, and exits judgement;
S8b: maximum probability and probability unlicensed in step S8a are more than or equal to the candidate of the suspected vehicles of threshold value
Region picture carries out the secondary discrimination, the method for the secondary discrimination as Target Photo are as follows:
The Target Photo is input in the model Y and is classified, classification standard is as follows:
If the maximum probability blocked completely, and probability is more than or equal to threshold value, then assert to exist and block behavior;
Otherwise assert that there is no block number plate behavior without exception.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811081354.8A CN109344886B (en) | 2018-09-17 | 2018-09-17 | Occlusion number plate distinguishing method based on convolutional neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811081354.8A CN109344886B (en) | 2018-09-17 | 2018-09-17 | Occlusion number plate distinguishing method based on convolutional neural network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109344886A true CN109344886A (en) | 2019-02-15 |
CN109344886B CN109344886B (en) | 2022-05-13 |
Family
ID=65305219
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811081354.8A Active CN109344886B (en) | 2018-09-17 | 2018-09-17 | Occlusion number plate distinguishing method based on convolutional neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109344886B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109919069A (en) * | 2019-02-27 | 2019-06-21 | 浙江浩腾电子科技股份有限公司 | Oversize vehicle analysis system based on deep learning |
CN110533042A (en) * | 2019-09-06 | 2019-12-03 | 北京慧智数据科技有限公司 | A kind of truck tail amplifying number recognition methods based on YOLO-V3 |
CN110909692A (en) * | 2019-11-27 | 2020-03-24 | 北京格灵深瞳信息技术有限公司 | Abnormal license plate recognition method and device, computer storage medium and electronic equipment |
CN111046956A (en) * | 2019-12-13 | 2020-04-21 | 苏州科达科技股份有限公司 | Occlusion image detection method and device, electronic equipment and storage medium |
CN111191604A (en) * | 2019-12-31 | 2020-05-22 | 上海眼控科技股份有限公司 | Method, device and storage medium for detecting integrity of license plate |
CN115083169A (en) * | 2022-06-14 | 2022-09-20 | 公安部交通管理科学研究所 | Method for discovering suspected vehicle imitating ambulance |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107301385A (en) * | 2017-06-09 | 2017-10-27 | 浙江宇视科技有限公司 | One kind blocks licence plate recognition method and device |
CN107341490A (en) * | 2017-07-17 | 2017-11-10 | 华中科技大学 | A kind of shielding automobile detection method and system based on convex closure analysis |
CN107977596A (en) * | 2016-10-25 | 2018-05-01 | 杭州海康威视数字技术股份有限公司 | A kind of car plate state identification method and device |
-
2018
- 2018-09-17 CN CN201811081354.8A patent/CN109344886B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107977596A (en) * | 2016-10-25 | 2018-05-01 | 杭州海康威视数字技术股份有限公司 | A kind of car plate state identification method and device |
CN107301385A (en) * | 2017-06-09 | 2017-10-27 | 浙江宇视科技有限公司 | One kind blocks licence plate recognition method and device |
CN107341490A (en) * | 2017-07-17 | 2017-11-10 | 华中科技大学 | A kind of shielding automobile detection method and system based on convex closure analysis |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109919069A (en) * | 2019-02-27 | 2019-06-21 | 浙江浩腾电子科技股份有限公司 | Oversize vehicle analysis system based on deep learning |
CN110533042A (en) * | 2019-09-06 | 2019-12-03 | 北京慧智数据科技有限公司 | A kind of truck tail amplifying number recognition methods based on YOLO-V3 |
CN110909692A (en) * | 2019-11-27 | 2020-03-24 | 北京格灵深瞳信息技术有限公司 | Abnormal license plate recognition method and device, computer storage medium and electronic equipment |
CN111046956A (en) * | 2019-12-13 | 2020-04-21 | 苏州科达科技股份有限公司 | Occlusion image detection method and device, electronic equipment and storage medium |
CN111191604A (en) * | 2019-12-31 | 2020-05-22 | 上海眼控科技股份有限公司 | Method, device and storage medium for detecting integrity of license plate |
CN115083169A (en) * | 2022-06-14 | 2022-09-20 | 公安部交通管理科学研究所 | Method for discovering suspected vehicle imitating ambulance |
CN115083169B (en) * | 2022-06-14 | 2023-07-18 | 公安部交通管理科学研究所 | Method for discovering suspected vehicles of counterfeit ambulances |
Also Published As
Publication number | Publication date |
---|---|
CN109344886B (en) | 2022-05-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109344886A (en) | Occlusion number plate distinguishing method based on convolutional neural network | |
TWI430212B (en) | Abnormal behavior detection system and method using automatic classification of multiple features | |
CN104951784B (en) | A kind of vehicle is unlicensed and license plate shading real-time detection method | |
CN105389567B (en) | Group abnormality detection method based on dense optical flow histogram | |
CN110969130B (en) | Driver dangerous action identification method and system based on YOLOV3 | |
CN105488453B (en) | A kind of driver based on image procossing does not fasten the safety belt detection recognition method | |
CN104077594B (en) | A kind of image-recognizing method and device | |
CN106203274A (en) | Pedestrian's real-time detecting system and method in a kind of video monitoring | |
Shan | Vehicle License Plate Recognition Based on Text-line Construction and Multilevel RBF Neural Network. | |
CN106204640A (en) | A kind of moving object detection system and method | |
CN110378179B (en) | Subway ticket evasion behavior detection method and system based on infrared thermal imaging | |
CN108960266A (en) | Image object detection method and device | |
CN107204114A (en) | A kind of recognition methods of vehicle abnormality behavior and device | |
CN102496058B (en) | Passenger flow density detection method | |
CN103871077B (en) | A kind of extraction method of key frame in road vehicles monitoring video | |
CN111680613B (en) | Method for detecting falling behavior of escalator passengers in real time | |
CN102163278B (en) | Illegal vehicle intruding detection method for bus lane | |
JP5931662B2 (en) | Road condition monitoring apparatus and road condition monitoring method | |
CN108804987B (en) | Door opening and closing state detection method and device and people flow detection system | |
CN104978567A (en) | Vehicle detection method based on scenario classification | |
CN102254428A (en) | Traffic jam detection method based on video processing | |
CN103390151A (en) | Face detection method and device | |
CN105956568A (en) | Abnormal behavior detecting and early warning method based on monitored object identification | |
CN105096344B (en) | Group behavior recognition methods and system based on CD motion features | |
CN107330027A (en) | A kind of Weakly supervised depth station caption detection method |
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 |