CN109949579A - A kind of illegal automatic auditing method that makes a dash across the red light based on deep learning - Google Patents
A kind of illegal automatic auditing method that makes a dash across the red light based on deep learning Download PDFInfo
- Publication number
- CN109949579A CN109949579A CN201811654645.1A CN201811654645A CN109949579A CN 109949579 A CN109949579 A CN 109949579A CN 201811654645 A CN201811654645 A CN 201811654645A CN 109949579 A CN109949579 A CN 109949579A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- line
- red light
- picture
- dash
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a kind of illegal automatic auditing method that makes a dash across the red light based on deep learning.The following steps are included: obtaining the original illegal picture that violation snap-shooting machine in front end uploads；Picture is subjected to cutting and rearrangement；Detect and identify the information of vehicles that needs are audited；Detect and identify the state of traffic lights；Original image is split, the necessary informations such as solid line, stop line, leading line, lane line, zebra stripes are partitioned into；Judge vehicle and stop line, leading line, the positional relationship of lane line etc.；The electronic police photo screening criteria that last basis is made a dash across the red light judges whether the original illegal picture is audited and passes through.This system realizes the automatic audit for the violation snap-shooting that makes a dash across the red light, and existing manual examination and verification method is substituted, has saved manpower, accelerates audit speed.
Description
Technical field
The present invention relates to the intelligent images such as target detection, attributive classification, scene cut to identify field technical field, especially relates to
And the artificial intelligence judgment technology field of the illegal electric police grasp shoot picture examination of automotive vehicle.
Background technique
Constantly improve with living standards of the people with the continuous social and economic development, Urban vehicles poputation rapidly increases
It is long.The quantity of the illegal electric police grasp shoot system of automotive vehicle also increases rapidly therewith.Traditional vehicle illegal electronic police
Capture picture examination mainly pass through manual examination and verification, workload is bigger, such as with special weather or road reformation electronics
In the case where police's cisco unity malfunction, a large amount of invalid candid photograph picture can be generated, which results in the workload of manual examination and verification is huge
Greatly.
How accurately and rapidly vehicle illegal electric police grasp shoot picture to be audited, at the same avoid desk checking at
This height, fatiguability, the easily drawbacks such as carelessness, are technical problems urgently to be solved.
Summary of the invention
The purpose of the present invention is: one kind is proposed for the violation snap-shooting automatic auditing method that makes a dash across the red light, and automatic picture of auditing is
It is no really to make a dash across the red light, with meet nowadays to motor vehicle make a dash across the red light efficiency that illegal electric police grasp shoot picture examination works,
The demand of accuracy rate.
The technical solution adopted by the present invention to solve the technical problems is:
A kind of illegal automatic auditing system that makes a dash across the red light based on deep learning, which comprises the following steps:
S1, the original illegal picture that violation snap-shooting machine in front end uploads is obtained；
S2, picture is subjected to cutting, 1-2 seconds evidence figures is divided between being cut into 3；
S3, detection simultaneously identify the information of vehicles that needs are audited, and first find out the vehicle for needing to audit with license plate recognition technology, so
Position of the vehicle in every evidence figure is found out using vehicle re-detection technology afterwards；
S4, the state for being detected based on deep learning and identifying traffic lights, comprehensive three evidence figures, the traffic light status is divided into
There is no a red light, left-hand rotation red light of keeping straight on, red light left-hand rotation green light of keeping straight on, green light left-hand rotation red light of keeping straight on, right-hand rotation traffic lights individually record；
S5, original image is split using deeplab-v2 partitioning algorithm, be partitioned into solid line, stop line, leading line,
The necessary informations such as lane line, zebra stripes；
S6, vehicle and stop line, leading line, the positional relationship of lane line etc. are judged based on image processing techniques；
The electronic police photo screening criteria that S7, basis are made a dash across the red light judges whether the original illegal picture is audited and passes through.
Further, the S3 finds out position of the vehicle in every evidence figure using vehicle re-detection (reid) technology and walks
It is rapid as follows:
S31, in training characteristics extraction module, in network, the last one 256 full articulamentum of dimension connects a classification layer, the layer
To classifying for different money vehicles, each classification possesses the same vehicle of different frame moment acquisition, and to the vehicle of all acquisitions
Carry out data enhancing.When trained penalty values loss is reduced to minimum, classification layer is cropped, takes out upper one 256 dimension Quan Lian
Layer is connect, 256 dimensional features obtained at this time can be good at characterizing the feature of the vehicle.
S32, GoogLenet Inception-V2 network is input to the vehicle that first figure navigates to, in the network
Input layer carries out padding to the vehicle of input, becomes the consistent image of length and width, extra part is with 0 pixel filling；Then
Up-sampling or down-sampling operation are carried out to pretreated image, unify resize at the image of 200*200 resolution ratio, finally
Obtain 256 dimensional features；
S33, vehicles input GoogLenet Inception-V2 network to be matched all to second figure, same to S32,
Obtain several 256 dimensional features；
S34, vehicles input GoogLenet Inception-V2 network to be matched all to third figure, same to S32,
Obtain several 256 dimensional features；
S35, cosine similarity is done with several 256 dimensional features in 256 dimensional features in S32 and S33, since feature mentions
256 dimensional features that modulus block extracts have been able to characterize the vehicle well, so can more show two using cosine is similar
Diversity factor between vehicle finally takes out 256 dimensional features corresponding to highest scoring；
S36, cosine similarity is done with several 256 dimensional features in 256 dimensional features of highest scoring in S33 and S34, taken out
256 dimensional features corresponding to highest scoring；
S37, several vehicles by detection algorithm have been detected due to second figure and third figure respectively, with above-mentioned calculation
Method finds the highest vehicle of similarity score, and taking out vehicle call number corresponding to highest scoring is the vehicle traced into.
Further, the traffic light status detecting step based on deep learning is as follows:
S41, the original big picture of input is divided into the splicing that small picture one by one has overlapping, overlapping area is traffic lights
The statistics maximum value of size；
S42, it small picture is sequentially input to SSD target detection network structure obtains the coordinate of target；
S43, obtained coordinates of targets is mapped to big figure above and merges the target of coincidence；
S44, it obtained target is input to traffic lights sorter network obtains the classification of traffic lights, sorter network uses
resnet18；
S45, the state that traffic lights are judged by the classification of each traffic lights；
Further, the S5 is split original image, be partitioned into solid line, stop line, leading line, lane line,
The necessary informations such as zebra stripes comprise the following steps that
S51, the picture for collecting application scenarios, and manually mark out solid line, stop line, leading line, lane line, zebra stripes etc.
Region；
S52, artificial mark is converted into label matrix, i.e., it is all pixels point in the closed polygon manually marked is corresponding
Label is set as 1, and the corresponding label of other pixels is set as 0；
S53, picture and the input deeplab-v2 partitioning algorithm training of corresponding label matrix, deeplab- v2 are divided
For algorithm using ResNet-34 as backbone network, psp_module and unet module uses skip layer as decoder
Low-dimensional minutia is introduced as prototype network structure.Use a*bce_loss+ b*lovasz_loss as final loss (0 <
=a, b≤1 are manually set), and introduce auxiliary loss aux_loss and be trained；
The good deeplab-v2 partitioning algorithm of S54, application training predicts input image pixels point classification, will belong to bus
The pixel coordinate set of road classification exports, to realize that road scene is divided；
Further, the vehicle and stop line based on image processing techniques, the positional relationship of leading line, lane line etc. are sentenced
It is disconnected that steps are as follows:
S61, the profile information for extracting solid line, stop line, leading line, lane line；
S62, straight line fitting is carried out using least square method, fits lane line and stop line；
S63, the minimum circumscribed rectangle for extracting leading line；
S64, the rectangle frame position for judging vehicle and lane straight line, stopping straight line, leading line boundary rectangle frame position
Relationship；
Further, the electronic police photo screening criteria that the basis is made a dash across the red light judges whether the original illegal picture is audited
By the way that steps are as follows:
The relationship of S71, the state for judging traffic lights and vehicle heading, i.e. through vehicles are left for red light of keeping straight on
Change trains or buses a corresponding left-hand rotation red light, right-turning vehicles correspond to right-hand rotation red light, do not turn left special lamp when, if keep straight on be red light if turn left
For red light, it is green light that green light of keeping straight on, which then turns left, and it is green light that right-hand rotation is defaulted if special lamp of not turning right；
S72, judge in first figure vehicle whether in stop line, judges whether vehicle in the second picture
Major part has passed over stop line, and whether third has obvious displacement with the second picture comparison vehicle, it is necessary to while meeting this
Three conditions are just regarded as making a dash across the red light.
The beneficial effects of the present invention are: make a dash across the red light violation snap-shooting auditing system present invention is mainly applied to automotive vehicle,
It realizes the automotive vehicle violation snap-shooting that makes a dash across the red light and audits automatically, has saved manpower, has improved efficiency, has in turn ensured audit work
That makees is just, openly.
Detailed description of the invention
Fig. 1: flow chart of the present invention
Fig. 2: schematic structural view of the invention
Fig. 3: traffic lights detection unit pattern splicing method schematic diagram of the present invention
Specific embodiment
Below in conjunction with attached drawing, the present invention will be further described.
Present invention is primarily based on rudimentary algorithm module, logic judgment module and regular judgment module.
As shown in Fig. 2, rudimentary algorithm module is divided into traffic lights detection taxon, vehicle detection unit, Car license recognition, vehicle
Re-detection unit and image segmentation unit.First picture is passed to traffic lights respectively and detects taxon, vehicle detection unit
Then vehicle detection result is passed to Car license recognition, vehicle re-detection unit and image segmentation unit.Wherein traffic lights detection is single
Member uses improved SSD target detection model, due to very small, the original detection side SSD of accounting of the traffic lights in whole picture figure
Method can not detect that a high-resolution pictures are divided into one by this one small target, a kind of method that the present invention uses picture mosaic
Open the small picture for having overlapping.As shown in Figure 3.
It is as follows that traffic lights detect disaggregated model acquisition methods:
S1, training data prepare: obtaining the image of different shooting conditions (such as illumination, angle)；
S2, data prediction: picture is cut into a sheet by a sheet small picture；
S2, data mark: traffic lights region is marked in the picture using rectangle frame, record the target area upper left corner and
The coordinate value of bottom right angle point；
S3, model training: using the training data marked, target detection model of the training based on deep learning；
The specific method of scene cut unit include: as shown in figure 3, segmentation module image is inputted into parted pattern first,
The classification of each pixel in image is obtained, to obtain leading for approximate location where target vehicle in lane line and first figure
Classification and region to line.
Scene cut model acquisition methods are as follows:
S1, training data prepare: obtaining the traffic intersection image of different shooting conditions (such as illumination, angle)；
S2, data mark: by lane line in image, zebra stripes and leading line carry out classification mark pixel-by-pixel；
S3, model training: using the training data marked, scene cut model of the training based on deep learning；
Implementation detailed process of the invention is as shown in Figure 1, a kind of illegal automatic audit system of making a dash across the red light based on deep learning
System, which comprises the following steps:
S1, the original illegal picture that violation snap-shooting machine in front end uploads is obtained；
S2, picture is subjected to cutting, is cut into 3 evidence figures；
S3, detection simultaneously identify the information of vehicles that needs are audited, and first find out the vehicle for needing to audit with license plate recognition technology, so
It integrates vehicle re-detection afterwards and license plate recognition technology finds out position of the vehicle in every evidence figure；
S4, the state for being detected based on deep learning and identifying traffic lights, comprehensive three evidence figures, the traffic light status is divided into
There is no a red light, left-hand rotation red light of keeping straight on, red light left-hand rotation green light of keeping straight on, green light left-hand rotation red light of keeping straight on, right-hand rotation traffic lights individually record；
S5, original image is split, is partitioned into the necessary letter such as solid line, stop line, leading line, lane line, zebra stripes
Breath；
S6, vehicle and stop line, leading line, the positional relationship of lane line etc. are judged based on image processing techniques；
The electronic police photo screening criteria that S7, basis are made a dash across the red light judges whether the original illegal picture is audited and passes through.
The advantages of basic principles and main features and this programme of this programme have been shown and described above.The technology of the industry
Personnel are it should be appreciated that this programme is not restricted to the described embodiments, and the above embodiments and description only describe this
The principle of scheme, under the premise of not departing from this programme spirit and scope, this programme be will also have various changes and improvements, these changes
Change and improvement is both fallen within the scope of claimed this programme.This programme be claimed range by appended claims and its
Equivalent thereof.
Claims (6)
1. a kind of illegal automatic auditing method that makes a dash across the red light based on deep learning, which comprises the following steps:
S1, the original illegal picture that violation snap-shooting machine in front end uploads is obtained；
S2, picture is subjected to cutting, is cut into the evidence figure that 3 interval times are respectively 1-2 seconds；
S3, detection simultaneously identify the information of vehicles that needs are audited, and first find out the vehicle for needing to audit with license plate recognition technology, then sharp
Position of the vehicle in every evidence figure is found out with vehicle re-detection technology；
S4, the state for being detected based on deep learning and identifying traffic lights, comprehensive three evidence figures, the traffic light status, which is divided into, not to be had
Red light, left-hand rotation red light of keeping straight on, red light left-hand rotation green light of keeping straight on, green light left-hand rotation red light of keeping straight on, right-hand rotation traffic lights individually record；
S5, original image is split using deeplab-v2 partitioning algorithm, is partitioned into solid line, stop line, leading line, lane
The necessary informations such as line, zebra stripes；
S6, vehicle and stop line, leading line, the positional relationship of lane line etc. are judged based on image processing techniques；
The electronic police photo screening criteria that S7, basis are made a dash across the red light judges whether the original illegal picture is audited and passes through.
2. a kind of illegal automatic auditing method that makes a dash across the red light based on deep learning as described in claim 1, which is characterized in that institute
S3 is stated using vehicle re-detection technology to find out position of the vehicle in every evidence figure steps are as follows:
S31, in training characteristics extraction module, in network, the last one 256 full articulamentum of dimension connects a classification layer, and the layer is not to
Classify with money vehicle, each classification possesses the same vehicle of different frame moment acquisition, and carries out the vehicle of all acquisitions
Data enhancing crops classification layer when trained penalty values loss is reduced to minimum, takes out the full articulamentum of upper one 256 dimension,
256 dimensional features obtained at this time can be good at characterizing the feature of the vehicle,
S32, GoogLenet Inception-V2 network is input to the vehicle that first figure navigates to, in the input of the network
Layer carries out padding to the vehicle of input, becomes the consistent image of length and width, extra part is with 0 pixel filling；Then to pre-
Treated, and image carries out up-sampling or down-sampling operation, and unified resize is finally obtained at the image of 200*200 resolution ratio
One 256 dimensional feature；
S33, GoogLenet Inception-V2 network, same to S32, if obtaining are inputted to all vehicles to be matched of second figure
Dry 256 dimensional features；
S34, GoogLenet Inception-V2 network, same to S32, if obtaining are inputted to all vehicles to be matched of third figure
Dry 256 dimensional features；
S35, cosine similarity is done with several 256 dimensional features in 256 dimensional features in S32 and S33, due to feature extraction mould
256 dimensional features that block extracts have been able to characterize the vehicle well, so can more show two cars using cosine is similar
Between diversity factor, finally take out highest scoring corresponding to 256 dimensional features；
S36, cosine similarity is made of several 256 dimensional features in 256 dimensional features of highest scoring in S33 and S34, take out score
256 dimensional features corresponding to highest；
S37, since second figure and third figure by detection algorithm have detected several vehicles respectively, looked for above-mentioned algorithm
To the highest vehicle of similarity score, taking out vehicle call number corresponding to highest scoring is the vehicle traced into.
3. a kind of illegal automatic auditing method that makes a dash across the red light based on deep learning as described in claim 1, which is characterized in that institute
It is as follows to state the traffic light status detecting step based on deep learning:
S41, the original big picture of input is divided into the splicing that small picture one by one has overlapping, overlapping area is traffic lights size
Statistics maximum value；
S42, it small picture is sequentially input to SSD target detection network structure obtains the coordinate of target；
S43, obtained coordinates of targets is mapped to big figure above and merges the target of coincidence；
S44, it obtained target is input to traffic lights sorter network obtains the classification of traffic lights, sorter network uses
resnet18；
S45, the state that traffic lights are judged by the classification of each traffic lights.
4. a kind of illegal automatic auditing method that makes a dash across the red light based on deep learning as described in claim 1, which is characterized in that institute
The S5's stated is split original image, is partitioned into the necessary informations packet such as solid line, stop line, leading line, lane line, zebra stripes
Include that steps are as follows:
S51, the picture for collecting application scenarios, and manually mark out the areas such as solid line, stop line, leading line, lane line, zebra stripes
Domain；
S52, artificial mark is converted into label matrix, i.e., by all pixels point corresponding label in the closed polygon manually marked
It is set as 1, the corresponding label of other pixels is set as 0；
S53, picture and the input deeplab-v2 partitioning algorithm training of corresponding label matrix, deeplab-v2 partitioning algorithm are adopted
Use ResNet-34 as backbone network, psp_module and unet module is as decoder, and it is low to use skip layer to introduce
Minutia is tieed up as prototype network structure, uses a*bce_loss+b*lovasz_loss as final loss, 0≤a, b
≤ 1, and introduce auxiliary loss aux_loss and be trained；
The good deeplab-v2 partitioning algorithm of S54, application training predicts input image pixels point classification, will belong to bus zone class
Other pixel coordinate set output, to realize that road scene is divided.
5. a kind of illegal automatic auditing method that makes a dash across the red light based on deep learning as described in claim 1, which is characterized in that institute
Vehicle and stop line based on image processing techniques are stated, the positional relationship judgment step of leading line, lane line etc. is as follows:
S61, the profile information for extracting solid line, stop line, leading line, lane line；
S62, straight line fitting is carried out using least square method, fits lane line and stop line；
S63, the minimum circumscribed rectangle for extracting leading line；
S64, the rectangle frame position for judging vehicle and lane straight line, stopping straight line, leading line boundary rectangle frame positional relationship.
6. a kind of illegal automatic auditing method that makes a dash across the red light based on deep learning as described in claim 1, which is characterized in that institute
It states and judges whether the original illegal picture is audited by the way that steps are as follows according to the electronic police photo screening criteria to make a dash across the red light:
The relationship of S71, the state for judging traffic lights and vehicle heading, i.e. through vehicles will be for red light of keeping straight on, left-hand rotation vehicles
Corresponding left-hand rotation red light, right-turning vehicles correspond to right-hand rotation red light, do not turn left special lamp when, it is red for turning left if straight trip is red light
Lamp, it is green light that straight trip green light, which then turns left, and it is green light that right-hand rotation is defaulted if special lamp of not turning right；
S72, whether vehicle is judged in first figure whether in stop line, judge in the second picture vehicle big portion
Divide and have passed over stop line, whether third has obvious displacement with the second picture comparison vehicle, it is necessary to while meeting these three
Condition is just regarded as making a dash across the red light.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811654645.1A CN109949579A (en) | 2018-12-31 | 2018-12-31 | A kind of illegal automatic auditing method that makes a dash across the red light based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811654645.1A CN109949579A (en) | 2018-12-31 | 2018-12-31 | A kind of illegal automatic auditing method that makes a dash across the red light based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109949579A true CN109949579A (en) | 2019-06-28 |
Family
ID=67006726
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811654645.1A Pending CN109949579A (en) | 2018-12-31 | 2018-12-31 | A kind of illegal automatic auditing method that makes a dash across the red light based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109949579A (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110459064A (en) * | 2019-09-19 | 2019-11-15 | 上海眼控科技股份有限公司 | Vehicle illegal behavioral value method, apparatus, computer equipment |
| CN110532903A (en) * | 2019-08-12 | 2019-12-03 | 浙江大华技术股份有限公司 | A kind of method and apparatus of traffic lights image procossing |
| CN110991221A (en) * | 2019-10-16 | 2020-04-10 | 合肥湛达智能科技有限公司 | Dynamic traffic red light running identification method based on deep learning |
| CN111292530A (en) * | 2020-02-04 | 2020-06-16 | 浙江大华技术股份有限公司 | Method, device, server and storage medium for processing violation pictures |
| CN111368774A (en) * | 2020-03-12 | 2020-07-03 | 北京以萨技术股份有限公司 | Waste film rollback method, system, terminal and medium based on traffic violation image |
| CN111401200A (en) * | 2020-03-10 | 2020-07-10 | 北京以萨技术股份有限公司 | Traffic violation picture processing method and device and readable storage medium |
| CN111462480A (en) * | 2020-02-28 | 2020-07-28 | 平安国际智慧城市科技股份有限公司 | Traffic image evidence verification method and device, computer equipment and storage medium |
| CN111462499A (en) * | 2020-03-26 | 2020-07-28 | 深圳极视角科技有限公司 | Method and device for detecting traffic violation |
| CN111583660A (en) * | 2020-05-22 | 2020-08-25 | 济南博观智能科技有限公司 | Vehicle steering behavior detection method, device, equipment and storage medium |
| CN111583661A (en) * | 2020-05-27 | 2020-08-25 | 魏保安 | Vehicle red light running detection method based on gun-ball linkage and DNN |
| CN111680580A (en) * | 2020-05-22 | 2020-09-18 | 北京格灵深瞳信息技术有限公司 | Red light running identification method and device, electronic equipment and storage medium |
| CN111768630A (en) * | 2019-12-24 | 2020-10-13 | 上海高德威智能交通***有限公司 | Violation waste image detection method and device and electronic equipment |
| CN111968378A (en) * | 2020-07-07 | 2020-11-20 | 浙江大华技术股份有限公司 | Motor vehicle red light running snapshot method and device, computer equipment and storage medium |
| CN113269060A (en) * | 2021-05-11 | 2021-08-17 | 北京易华录信息技术股份有限公司 | Vehicle illegal behavior review method and device, electronic equipment and storage medium |
| CN114255598A (en) * | 2021-12-13 | 2022-03-29 | 以萨技术股份有限公司 | Vehicle illegal data processing method, system and storage medium |
| CN114299414A (en) * | 2021-11-30 | 2022-04-08 | 无锡数据湖信息技术有限公司 | Deep learning-based vehicle red light running identification and determination method |
| CN115482659A (en) * | 2022-08-18 | 2022-12-16 | 浙江工商大学 | Intelligent agent autonomous decision-making method based on deep reinforcement learning |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101916383A (en) * | 2010-08-25 | 2010-12-15 | 浙江师范大学 | Vehicle detecting, tracking and identifying system based on multi-camera |
| JP2011123613A (en) * | 2009-12-09 | 2011-06-23 | Fuji Heavy Ind Ltd | Stop line recognition device |
| CN102110369A (en) * | 2010-12-21 | 2011-06-29 | 汉王科技股份有限公司 | Jaywalking snapshot method and device |
| CN201984638U (en) * | 2011-01-25 | 2011-09-21 | 上海市金山区青少年活动中心 | Control device for managing vehicles running red light |
| CN102521964A (en) * | 2011-11-28 | 2012-06-27 | 重庆警官职业学院 | Traffic violation processing method based on cloud computing |
| CN103345618A (en) * | 2013-06-21 | 2013-10-09 | 银江股份有限公司 | Traffic violation detection method based on video technology |
| CN103778786A (en) * | 2013-12-17 | 2014-05-07 | 东莞中国科学院云计算产业技术创新与育成中心 | Traffic violation detection method based on significant vehicle part model |
| CN104050447A (en) * | 2014-06-05 | 2014-09-17 | 奇瑞汽车股份有限公司 | Traffic light identification method and device |
| CN104715615A (en) * | 2015-04-08 | 2015-06-17 | 姜翠英 | Electronic violation recognizing platform in traffic intersection |
| CN105809965A (en) * | 2016-05-25 | 2016-07-27 | 成都联众智科技有限公司 | Automatic processing system for road monitoring images |
| CN106327453A (en) * | 2015-06-30 | 2017-01-11 | 北京金山安全软件有限公司 | Method for splicing picture resources and picture resource splicing device |
| CN107134144A (en) * | 2017-04-27 | 2017-09-05 | 武汉理工大学 | A kind of vehicle checking method for traffic monitoring |
| CN107301777A (en) * | 2016-11-25 | 2017-10-27 | 上海炬宏信息技术有限公司 | Vehicle peccancy lane change detection method based on video detection technology |
| US10037691B1 (en) * | 2017-03-31 | 2018-07-31 | International Business Machines Corporation | Behavioral based traffic infraction detection and analysis system |
| CN108765386A (en) * | 2018-05-16 | 2018-11-06 | 中铁科学技术开发公司 | A kind of tunnel slot detection method, device, electronic equipment and storage medium |
-
2018
- 2018-12-31 CN CN201811654645.1A patent/CN109949579A/en active Pending
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011123613A (en) * | 2009-12-09 | 2011-06-23 | Fuji Heavy Ind Ltd | Stop line recognition device |
| CN101916383A (en) * | 2010-08-25 | 2010-12-15 | 浙江师范大学 | Vehicle detecting, tracking and identifying system based on multi-camera |
| CN102110369A (en) * | 2010-12-21 | 2011-06-29 | 汉王科技股份有限公司 | Jaywalking snapshot method and device |
| CN201984638U (en) * | 2011-01-25 | 2011-09-21 | 上海市金山区青少年活动中心 | Control device for managing vehicles running red light |
| CN102521964A (en) * | 2011-11-28 | 2012-06-27 | 重庆警官职业学院 | Traffic violation processing method based on cloud computing |
| CN103345618A (en) * | 2013-06-21 | 2013-10-09 | 银江股份有限公司 | Traffic violation detection method based on video technology |
| CN103778786A (en) * | 2013-12-17 | 2014-05-07 | 东莞中国科学院云计算产业技术创新与育成中心 | Traffic violation detection method based on significant vehicle part model |
| CN104050447A (en) * | 2014-06-05 | 2014-09-17 | 奇瑞汽车股份有限公司 | Traffic light identification method and device |
| CN104715615A (en) * | 2015-04-08 | 2015-06-17 | 姜翠英 | Electronic violation recognizing platform in traffic intersection |
| CN105448096A (en) * | 2015-04-08 | 2016-03-30 | 姜翠英 | Law violation electronic identification platform positioned at crossroad |
| CN106327453A (en) * | 2015-06-30 | 2017-01-11 | 北京金山安全软件有限公司 | Method for splicing picture resources and picture resource splicing device |
| CN105809965A (en) * | 2016-05-25 | 2016-07-27 | 成都联众智科技有限公司 | Automatic processing system for road monitoring images |
| CN107301777A (en) * | 2016-11-25 | 2017-10-27 | 上海炬宏信息技术有限公司 | Vehicle peccancy lane change detection method based on video detection technology |
| US10037691B1 (en) * | 2017-03-31 | 2018-07-31 | International Business Machines Corporation | Behavioral based traffic infraction detection and analysis system |
| CN107134144A (en) * | 2017-04-27 | 2017-09-05 | 武汉理工大学 | A kind of vehicle checking method for traffic monitoring |
| CN108765386A (en) * | 2018-05-16 | 2018-11-06 | 中铁科学技术开发公司 | A kind of tunnel slot detection method, device, electronic equipment and storage medium |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110532903B (en) * | 2019-08-12 | 2022-02-22 | 浙江大华技术股份有限公司 | Traffic light image processing method and equipment |
| CN110532903A (en) * | 2019-08-12 | 2019-12-03 | 浙江大华技术股份有限公司 | A kind of method and apparatus of traffic lights image procossing |
| CN110459064A (en) * | 2019-09-19 | 2019-11-15 | 上海眼控科技股份有限公司 | Vehicle illegal behavioral value method, apparatus, computer equipment |
| CN110991221A (en) * | 2019-10-16 | 2020-04-10 | 合肥湛达智能科技有限公司 | Dynamic traffic red light running identification method based on deep learning |
| CN110991221B (en) * | 2019-10-16 | 2024-02-27 | 合肥湛达智能科技有限公司 | Dynamic traffic red light running recognition method based on deep learning |
| CN111768630A (en) * | 2019-12-24 | 2020-10-13 | 上海高德威智能交通***有限公司 | Violation waste image detection method and device and electronic equipment |
| CN111292530A (en) * | 2020-02-04 | 2020-06-16 | 浙江大华技术股份有限公司 | Method, device, server and storage medium for processing violation pictures |
| CN111462480A (en) * | 2020-02-28 | 2020-07-28 | 平安国际智慧城市科技股份有限公司 | Traffic image evidence verification method and device, computer equipment and storage medium |
| CN111462480B (en) * | 2020-02-28 | 2023-02-07 | 平安国际智慧城市科技股份有限公司 | Traffic image evidence verification method and device, computer equipment and storage medium |
| CN111401200A (en) * | 2020-03-10 | 2020-07-10 | 北京以萨技术股份有限公司 | Traffic violation picture processing method and device and readable storage medium |
| CN111368774A (en) * | 2020-03-12 | 2020-07-03 | 北京以萨技术股份有限公司 | Waste film rollback method, system, terminal and medium based on traffic violation image |
| CN111462499A (en) * | 2020-03-26 | 2020-07-28 | 深圳极视角科技有限公司 | Method and device for detecting traffic violation |
| CN111680580A (en) * | 2020-05-22 | 2020-09-18 | 北京格灵深瞳信息技术有限公司 | Red light running identification method and device, electronic equipment and storage medium |
| CN111583660A (en) * | 2020-05-22 | 2020-08-25 | 济南博观智能科技有限公司 | Vehicle steering behavior detection method, device, equipment and storage medium |
| CN111583661A (en) * | 2020-05-27 | 2020-08-25 | 魏保安 | Vehicle red light running detection method based on gun-ball linkage and DNN |
| CN111968378A (en) * | 2020-07-07 | 2020-11-20 | 浙江大华技术股份有限公司 | Motor vehicle red light running snapshot method and device, computer equipment and storage medium |
| CN113269060A (en) * | 2021-05-11 | 2021-08-17 | 北京易华录信息技术股份有限公司 | Vehicle illegal behavior review method and device, electronic equipment and storage medium |
| CN113269060B (en) * | 2021-05-11 | 2024-02-13 | 北京易华录信息技术股份有限公司 | Vehicle illegal behavior review method and device, electronic equipment and storage medium |
| CN114299414A (en) * | 2021-11-30 | 2022-04-08 | 无锡数据湖信息技术有限公司 | Deep learning-based vehicle red light running identification and determination method |
| CN114299414B (en) * | 2021-11-30 | 2023-09-15 | 无锡数据湖信息技术有限公司 | Vehicle red light running recognition and judgment method based on deep learning |
| CN114255598A (en) * | 2021-12-13 | 2022-03-29 | 以萨技术股份有限公司 | Vehicle illegal data processing method, system and storage medium |
| CN115482659A (en) * | 2022-08-18 | 2022-12-16 | 浙江工商大学 | Intelligent agent autonomous decision-making method based on deep reinforcement learning |
| CN115482659B (en) * | 2022-08-18 | 2023-09-19 | 浙江工商大学 | Intelligent autonomous decision-making method based on deep reinforcement learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109949579A (en) | A kind of illegal automatic auditing method that makes a dash across the red light based on deep learning | |
| CN109948418A (en) | A kind of illegal automatic auditing method of violation guiding based on deep learning | |
| CN109637151A (en) | A kind of recognition methods that highway Emergency Vehicle Lane is driven against traffic regulations | |
| CN110197589B (en) | Deep learning-based red light violation detection method | |
| CN104751634B (en) | The integrated application method of freeway tunnel driving image acquisition information | |
| CN111161543A (en) | Automatic snapshot method and system for bus front violation behavior based on image recognition | |
| CN105702048B (en) | Highway front truck illegal road occupation identifying system based on automobile data recorder and method | |
| CN102722704B (en) | Method and system for recognizing vehicle license plate by integrating video dynamic tracking | |
| CN103824452B (en) | A kind of peccancy parking detector based on panoramic vision of lightweight | |
| CN102867417B (en) | Taxi anti-forgery system and taxi anti-forgery method | |
| CN109948416A (en) | A kind of illegal occupancy bus zone automatic auditing method based on deep learning | |
| CN101383003B (en) | Real-time precise recognition method for vehicle number board | |
| CN102509090B (en) | A kind of vehicle feature recognition device based on public safety video image in sky net engineering | |
| CN106373426A (en) | Computer vision-based parking space and illegal lane occupying parking monitoring method | |
| CN110689724B (en) | Automatic motor vehicle zebra crossing present pedestrian auditing method based on deep learning | |
| CN102708378A (en) | Method for diagnosing fault of intelligent traffic capturing equipment based on image abnormal characteristic | |
| CN103186982A (en) | Intelligent access system for public security | |
| CN103279756A (en) | Vehicle detecting analysis system and detecting analysis method thereof based on integrated classifier | |
| CN103646544B (en) | Based on the vehicle behavioural analysis recognition methods of The Cloud Terrace and camera apparatus | |
| CN103390167A (en) | Multi-characteristic layered traffic sign identification method | |
| CN202472943U (en) | Vehicle violation detecting system based on high definition video technology | |
| CN106909923A (en) | A kind of vehicles peccancy processing system | |
| CN106919939B (en) | A kind of traffic signboard tracks and identifies method and system | |
| CN109948417A (en) | A kind of vehicle based on deep learning drives in the wrong direction illegal automatic auditing method | |
| CN106503640A (en) | A kind of detection method for taking bus zone |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190628 |
|
| WD01 | Invention patent application deemed withdrawn after publication |