CN114049771A - Bimodal-based traffic anomaly detection method and system and storage medium - Google Patents
Bimodal-based traffic anomaly detection method and system and storage medium Download PDFInfo
- Publication number
- CN114049771A CN114049771A CN202210029611.3A CN202210029611A CN114049771A CN 114049771 A CN114049771 A CN 114049771A CN 202210029611 A CN202210029611 A CN 202210029611A CN 114049771 A CN114049771 A CN 114049771A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- vehicles
- video stream
- static
- dynamic
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 89
- 230000002902 bimodal effect Effects 0.000 title claims abstract description 37
- 230000003068 static effect Effects 0.000 claims abstract description 102
- 230000002159 abnormal effect Effects 0.000 claims abstract description 93
- 230000009977 dual effect Effects 0.000 claims abstract description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 41
- 230000015654 memory Effects 0.000 claims description 15
- 230000011218 segmentation Effects 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 10
- 239000013598 vector Substances 0.000 claims description 10
- 238000007621 cluster analysis Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 7
- 238000010586 diagram Methods 0.000 claims description 6
- 230000002708 enhancing effect Effects 0.000 claims description 5
- 230000002401 inhibitory effect Effects 0.000 claims description 5
- 238000011176 pooling Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 230000002547 anomalous effect Effects 0.000 claims 1
- 238000000034 method Methods 0.000 description 18
- 230000005856 abnormality Effects 0.000 description 12
- 239000011159 matrix material Substances 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000004927 fusion Effects 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000007723 transport mechanism Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
- G08G1/0175—Detecting movement of traffic to be counted or controlled identifying vehicles by photographing vehicles, e.g. when violating traffic rules
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (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)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a traffic anomaly detection method, a system and a storage medium based on dual modes, which comprise the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data. The bimodal-based traffic anomaly detection method provided by the invention can learn from static and dynamic vehicle motion modes, detect various road traffic anomaly events in a real scene, and effectively detect the occurrence of traffic anomalies.
Description
Technical Field
The invention relates to the technical field of traffic anomaly detection, in particular to a bimodal-based traffic anomaly detection method, a bimodal-based traffic anomaly detection system and a storage medium.
Background
With the higher and higher holding amount of household vehicles, the automobile traveling is also a very common way. Therefore, urban road conditions are of great public concern, and severe road conditions may cause great loss to social economy and threaten the personal safety of drivers. It is important to monitor road conditions by widely deploying traffic cameras and to develop a method for detecting abnormal accidents in real time using computer vision technology. Many benefits and conveniences can be brought about by deploying these algorithmic devices. When the abnormity happens, the traffic management personnel can be immediately informed to handle, and the monitored road condition information can provide guidance for planned travel of the vehicle.
Normally, the vehicle will remain in motion on the road except for certain normal conditions (e.g., waiting for traffic lights). Therefore, the static vehicle has a higher probability of occurrence of an abnormal event. Typically, most exceptional events in road traffic result in parking. Such as a car stall, traffic accident or car jam. For the task of detecting traffic abnormality, it is critical to find out the time and place where the abnormality occurs. Therefore, it is necessary to design a method for detecting various traffic abnormal events in a real scene.
However, designing image detection algorithms to detect anomalies in road traffic is a very challenging task. The main reason is that the movement pattern of vehicles on the road is often very complex, and different abnormal events may show very complex behavior. Meanwhile, compared with normal events, abnormal events are seldom generated, many abnormal detection works in the current monitoring video can only be used for detecting specific abnormal events, and the abnormal events on roads are complex and various.
Disclosure of Invention
The invention aims to overcome the technical defects, provides a traffic abnormity detection method, a traffic abnormity detection system and a storage medium based on a dual mode, and solves the technical problem that complex abnormal events on roads cannot be intelligently processed in the prior art.
In order to achieve the above technical objective, a first aspect of the present invention provides a traffic anomaly detection method based on dual modes, including the following steps:
acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames;
detecting and identifying static vehicles in picture frames of the traffic video stream data;
detecting dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks;
comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of an abnormal event according to the similarity and the time data.
Compared with the prior art, the bimodal-based traffic anomaly detection method provided by the invention has the beneficial effects that:
the traffic anomaly detection method based on the dual modes comprises the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in terms of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of an abnormal event according to the similarity and the time data. The bimodal-based traffic anomaly detection method provided by the invention can learn from static and dynamic vehicle motion modes, detect various road traffic anomaly events in a real scene, and effectively detect the occurrence of traffic anomalies.
According to some embodiments of the invention, the detecting and identifying the static vehicle in the picture frame of the traffic video stream data comprises the steps of:
calculating the weighted sum of input frames in the traffic video stream data, enhancing the static part of a picture, inhibiting the moving part of the picture, and extracting all picture frames containing static vehicle data in the traffic video stream data;
detecting a static vehicle in the picture frame using a yolo algorithm.
According to some embodiments of the invention, the calculation formula of the weighted sum of the input frames in the traffic video stream data is:
wherein,
an ith frame representing a video;
is the average from frame 1 to frame i;
is the weight value of the image frame.
According to some embodiments of the invention, said detecting a static vehicle in said picture frame using the yolo algorithm comprises the steps of:
preprocessing the picture frame and setting the picture frame to be a preset size;
the information of the previous picture frame is reserved through a residual error unit, and the dimensionality of the tensor is spliced and expanded through a tensor mode;
extracting image features of the picture frame through a convolutional layer to detect the static vehicle in the picture frame.
According to some embodiments of the invention, the detecting the dynamic vehicle in the traffic video stream data, tracking and identifying the dynamic vehicle and marking the abnormal track vehicle comprises the following steps:
segmenting dynamic vehicles in the traffic video stream data by using a semantic segmentation algorithm;
tracking the dynamic vehicle using a target tracking algorithm;
and calculating the track of the dynamic vehicle, carrying out cluster analysis on the track to obtain a main motion mode of the dynamic vehicle, and marking the dynamic vehicle deviating from the main motion mode as an abnormal track vehicle.
According to some embodiments of the invention, the segmenting of the dynamic vehicles in the traffic video stream data using a semantic segmentation algorithm comprises the steps of:
preprocessing the picture frame;
extracting a characteristic diagram of the picture frame through a convolution layer of a mask-rcnn network;
inputting the feature map into an RPN network, and calculating a candidate region of the picture frame;
pooling the feature maps of the candidate regions to a fixed size by a ROIAlign operation;
and in a mask-rcndn classification regression network, classifying the feature graph and regressing a target frame, and generating a mask through full-connection layer detection to segment the dynamic vehicle.
According to some embodiments of the invention, the tracking the dynamic vehicle using a target tracking algorithm comprises the steps of:
predicting to obtain the vehicle state information of the current frame according to the speed and position information of the dynamic vehicle in the previous frame and based on a Kalman filtering algorithm;
estimating the real state of the current dynamic vehicle according to the detection result and the prediction result of the current frame;
extracting a feature vector of the dynamic vehicle in a target frame by using a feature extraction network;
calculating the cosine distance of the characteristic vectors of the front frame and the rear frame as a quantized value of the similarity, matching the vehicles in the detection frames of the front frame and the rear frame, and distributing the same serial number to the dynamic vehicles with the matching degree above a preset threshold value.
According to some embodiments of the invention, the calculating the similarity between the static vehicle and the vehicle with the abnormal track comprises:
and comparing the cosine similarity of the characteristic values of the static vehicle and the abnormal track vehicle, and when the cosine similarity is higher than a threshold value, determining that the static vehicle and the abnormal track vehicle are the same vehicle.
In a second aspect, the present invention also provides a traffic anomaly detection system based on dual modes, including: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the bimodal-based traffic anomaly detection method according to any one of the first aspect when executing the computer program.
In a third aspect, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions for causing a computer to execute the bimodal-based traffic anomaly detection method according to any one of the first aspect.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which the abstract is to be fully consistent with one of the figures of the specification:
FIG. 1 is a flow chart of a bimodal-based traffic anomaly detection method according to an embodiment of the present invention;
fig. 2 is a flowchart of a bimodal-based traffic anomaly detection method according to another embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
It should be noted that although functional block divisions are provided in the system drawings and logical orders are shown in the flowcharts, in some cases, the steps shown and described may be performed in different orders than the block divisions in the systems or in the flowcharts. The terms first, second and the like in the description and in the claims, and the drawings described above, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.
The invention provides a bimodal-based traffic anomaly detection method, which can learn from static and dynamic vehicle motion modes, detect various road traffic anomaly events in a real scene and effectively detect the occurrence of traffic anomalies.
The embodiments of the present invention will be further explained with reference to the drawings.
Referring to fig. 1, fig. 1 is a flowchart of a bimodal-based traffic anomaly detection method according to an embodiment of the present invention.
The bimodal-based traffic abnormality detection method includes, but is not limited to, steps S110 to S140.
Step S110, traffic video stream data is obtained, and the video stream data comprises time data and picture frames;
step S120, detecting and identifying static vehicles in the picture frames of the traffic video stream data;
step S130, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks;
step S140, comparing the characteristics of the static vehicle and the abnormal track vehicle in terms of brightness, contrast and structure, calculating the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data.
It should be noted that step S120 and step S130 may be performed simultaneously, and the execution sequence of step S120 and step S130 is not limited in the above embodiment.
In one embodiment, the bimodal-based traffic anomaly detection method comprises the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data. The bimodal-based traffic anomaly detection method provided by the invention can learn from static and dynamic vehicle motion modes, detect various road traffic anomaly events in a real scene, and effectively detect the occurrence of traffic anomalies.
In the embodiment, video stream data is pulled and decoded by accessing a traffic monitoring camera ip, static vehicles in a video are extracted through motion analysis, vehicles which are abnormally stopped are detected by using a yolo algorithm and a classifier, the moving vehicles are segmented by mask-rcnn, target vehicles are tracked by depsort, and vehicles with abnormal tracks are detected by cluster analysis. And matching the results of the static mode detection method and the motion mode detection method with vehicles through SSIM, and finally outputting the fusion result of the abnormal events.
In one embodiment, the bimodal-based traffic anomaly detection method comprises the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data.
Detecting and identifying static vehicles in a picture frame of traffic video stream data, comprising the steps of: calculating the weighted sum of input frames in traffic video stream data, enhancing a static part of a picture, inhibiting a moving part in the picture, and extracting all picture frames containing static vehicle data in the traffic video stream data; a yolo algorithm is used to detect static vehicles in the picture frames. The bimodal-based traffic anomaly detection method provided by the invention can learn from static and dynamic vehicle motion modes, detect various road traffic anomaly events in a real scene, and effectively detect the occurrence of traffic anomalies.
In one embodiment, the bimodal-based traffic anomaly detection method comprises the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data.
Detecting and identifying static vehicles in a picture frame of traffic video stream data, comprising the steps of: calculating the weighted sum of input frames in traffic video stream data, enhancing a static part of a picture, inhibiting a moving part in the picture, and extracting all picture frames containing static vehicle data in the traffic video stream data; a yolo algorithm is used to detect static vehicles in the picture frames. Detecting a static vehicle in a picture frame using the yolo algorithm, comprising the steps of: preprocessing a picture frame, and setting the picture frame to be a preset size; the information of the previous picture frame is reserved through a residual error unit, and the dimensionality of the tensor is spliced and expanded through a tensor mode; and extracting image characteristics of the picture frame through the convolution layer to detect the static vehicle in the picture frame.
In one embodiment, the bimodal-based traffic anomaly detection method comprises the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data. The method for detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks comprises the following steps: segmenting dynamic vehicles in traffic video stream data by using a semantic segmentation algorithm; tracking the dynamic vehicle by using a target tracking algorithm; calculating the track of the dynamic vehicle, carrying out cluster analysis on the track to obtain a main motion mode of the dynamic vehicle, and marking the dynamic vehicle deviating from the main motion mode as an abnormal track vehicle.
In one embodiment, the bimodal-based traffic anomaly detection method comprises the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data. The method for acquiring traffic video stream data comprises the following steps: and accessing an intranet of the camera through the embedded industrial personal computer, accessing multiple ip addresses of the camera, and acquiring traffic video stream data of the traffic detection camera.
In one embodiment, the bimodal-based traffic anomaly detection method comprises the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data.
The method for detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks comprises the following steps: segmenting dynamic vehicles in traffic video stream data by using a semantic segmentation algorithm; tracking the dynamic vehicle by using a target tracking algorithm; calculating the track of the dynamic vehicle, carrying out cluster analysis on the track to obtain a main motion mode of the dynamic vehicle, and marking the dynamic vehicle deviating from the main motion mode as an abnormal track vehicle.
The semantic segmentation algorithm is used for segmenting the dynamic vehicles in the traffic video stream data, and the semantic segmentation algorithm comprises the following steps: preprocessing a picture frame; extracting a characteristic diagram of the picture frame through a convolution layer of a mask-rcnn network; inputting the feature map into an RPN network, and calculating a candidate area of a picture frame; pooling the feature maps of the candidate regions to a fixed size by a roiign operation; in a mask-rcndn classification regression network, classification and regression of a target frame are carried out on the feature diagram, a mask is generated through full-connection layer detection, and the dynamic vehicle is segmented.
In one embodiment, the bimodal-based traffic anomaly detection method comprises the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data.
The method for detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks comprises the following steps: segmenting dynamic vehicles in traffic video stream data by using a semantic segmentation algorithm; tracking the dynamic vehicle by using a target tracking algorithm; calculating the track of the dynamic vehicle, carrying out cluster analysis on the track to obtain a main motion mode of the dynamic vehicle, and marking the dynamic vehicle deviating from the main motion mode as an abnormal track vehicle. The semantic segmentation algorithm is used for segmenting the dynamic vehicles in the traffic video stream data, and the semantic segmentation algorithm comprises the following steps: preprocessing a picture frame; extracting a characteristic diagram of the picture frame through a convolution layer of a mask-rcnn network; inputting the feature map into an RPN network, and calculating a candidate area of a picture frame; pooling the feature maps of the candidate regions to a fixed size by a roiign operation; in a mask-rcndn classification regression network, classification and regression of a target frame are carried out on the feature diagram, a mask is generated through full-connection layer detection, and the dynamic vehicle is segmented.
Tracking a dynamic vehicle using a target tracking algorithm, comprising the steps of: predicting to obtain the vehicle state information of the current frame according to the speed and position information of the dynamic vehicle in the previous frame and based on a Kalman filtering algorithm; estimating the real state of the current dynamic vehicle according to the combination of the detection result and the prediction result of the current frame; extracting a feature vector of the dynamic vehicle in the target frame by using a feature extraction network; calculating the cosine distance of the characteristic vectors of the front frame and the rear frame as a quantized value of the similarity, matching the vehicles in the detection frames of the front frame and the rear frame, and distributing the same serial number to the dynamic vehicles with the matching degree above a preset threshold value.
In one embodiment, the bimodal-based traffic anomaly detection method comprises the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data. The method for calculating the similarity of the static vehicle and the abnormal track vehicle comprises the following steps: and comparing the cosine similarity of the characteristic values of the static vehicle and the abnormal track vehicle, and when the cosine similarity is higher than a threshold value, determining that the static vehicle and the abnormal track vehicle are the same vehicle.
In one embodiment, the bimodal-based traffic anomaly detection method comprises the following steps: firstly, acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames; then, detecting and identifying static vehicles in the picture frames of the traffic video stream data; secondly, detecting the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks; and finally, comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of the abnormal event according to the similarity and the time data.
Detecting and identifying static vehicles in a picture frame of traffic video stream data, comprising the steps of: calculating the weighted sum of input frames in traffic video stream data, enhancing a static part of a picture, inhibiting a moving part in the picture, and extracting all picture frames containing static vehicle data in the traffic video stream data; a yolo algorithm is used to detect static vehicles in the picture frames. The bimodal-based traffic anomaly detection method provided by the invention can learn from static and dynamic vehicle motion modes, detect various road traffic anomaly events in a real scene, and effectively detect the occurrence of traffic anomalies.
The formula for computing the weighted sum of the input frames in the traffic video stream data is:
wherein,
an ith frame representing a video;
is the average from frame 1 to frame i;
is the weight value of the image frame. By regulating
The update speed of the average image can be controlled, and after all videos are processed, the static vehicle images in all frames can be extracted.
In one embodiment, the traffic anomaly detection method based on the bimodal analysis mainly comprises the steps of traffic video stream data acquisition, anomaly detection of a vehicle static mode, anomaly detection of a vehicle motion mode and bimodal fusion based on SSIM. Referring to fig. 2, fig. 2 is a flowchart of a bimodal-based traffic anomaly detection method according to another embodiment of the present invention. The specific operation comprises the following steps:
1. acquiring traffic video stream data:
the embedded industrial personal computer is accessed into a camera intranet, a camera ip address is accessed to pull a video stream of the traffic detection camera, and the video stream contains attributes such as a timestamp and a frame number. The pulled video stream is analyzed in real time by a program deployed on an industrial personal computer, and in the normal driving process, traffic flow data can be sent to an abnormality detection module to detect traffic abnormality in real time.
2. Abnormality detection of vehicle static mode:
the abnormality detection based on the static vehicle is mainly to find the abnormality by a two-step method. And (4) motion analysis, extracting static vehicles in the video vehicles, and detecting and identifying abnormally stopped vehicles by using a yolo algorithm and a two-classifier. The specific process is as follows:
(1) background modeling
The static part of the picture is enhanced and the moving part of the picture is suppressed by continuously calculating the weighted sum of the input frames in the whole video, and the weighting formula is as follows:
wherein,
an ith frame representing a video;
is the average from frame 1 to frame i;
is the weight value of the image frame.
By regulating
The update speed of the average image can be controlled, and after all videos are processed, the static vehicle images in all frames can be extracted.
(2) Detection and identification
The yolo algorithm is used to detect static vehicles in the frame. The yolo algorithm sets the image size to 416 x 416 through image preprocessing, extracts image features through a convolutional layer, retains front layer information through a residual error unit, avoids gradient disappearance, expands the dimensionality of a tensor through tensor splicing, and increases the detection performance of a model through an up-sampling unit. The yolo algorithm has better detection precision for background, and can detect static vehicles and pseudo-static vehicles in an average image.
(3) False detection deletion
In the detection result of the average image, many noises such as camera shake exist, and the yolo algorithm is mistaken for a static vehicle. The detection results contain some false positives from the background image, using a two-classifier to distinguish between the vehicle and the background area. The sample of the classifier can be obtained by collecting monitoring traffic videos, the background image patch is cut randomly to obtain a negative sample, the automobile image is cut randomly to obtain a positive sample, and a two-classifier network is trained. The network model can distinguish real static vehicles and background in yolo detection results more quickly.
3. Abnormality detection of vehicle motion modality:
after the collected traffic video stream is decoded, a target vehicle is segmented by mask-rcnn for each frame of image, the detected vehicle is tracked by depsort, and a vehicle with abnormal speed is detected. The specific process is as follows:
(1) target object segmentation
Preprocessing a single frame image obtained by decoding, extracting the characteristics of a video image through a convolution layer of a mask-rcnn network after preprocessing is finished, inputting the characteristic image into an RPN network, calculating a candidate region of the image, pooling the characteristic image of the candidate region into a fixed size through ROIAlign operation, classifying the characteristic image and regressing a target frame in a mask-rcndn classification regression network, detecting and generating a mask through a full connection layer, and segmenting out a target vehicle.
(2) Target tracking
And tracking the target vehicle segmented by the mask-rcnn by using a deepsort algorithm. Firstly, using a Kalman filtering algorithm, predicting the vehicle state information of a current frame according to the speed and position information of a target in a previous frame, and estimating the real state of the current target according to the detection result and the prediction result of the current frame.
Extracting the feature vectors of the objects in the target frame through the trained feature extraction network, calculating the cosine distance of the feature vectors of the front frame and the rear frame as the quantization value of the similarity, matching the vehicles in the detection frames of the front frame and the rear frame by using the Hungarian algorithm, and allocating the same number to the target objects with high matching degree.
(3) Anomaly detection
And calculating the track of the target vehicle, carrying out cluster analysis on the track, finding out a main motion mode, and regarding the vehicle deviating from the main motion mode as an abnormal vehicle. And using a clustering method based on similar information for the normal track of the road.
First a set of vectors is determined
Wherein
And
is similar:
learning a distance metric by mahalanobis distance formula:
to ensure that the distance metric given above satisfies non-negativity, the matrix
Must be semi-positive.
For known similar data, which require their sum of squared distances to be as small as possible, the distance metric parameters are learned by establishing the following objective function:
simultaneously, the following constraint conditions are satisfied:
matrix array
Is equivalent to distance learning, so it turns into the following optimization problem:
in order to increase the value of equation (7), a gradient-increasing method is applied to equation (7):
then sequentially arranging the matrixes
Projection to a set
First step of projection to
The problem of minimizing a quadratic objective function under the condition of independent linear constraint can be solved by directly solving a linear equation set;
second step of projection to
First, a contract matrix of a diagonal matrix is found
Wherein,
is that
The characteristic values of (a) are combined into a diagonal matrix,
is that
The corresponding feature vector.
Then get it again
Wherein,
the distance learning method is used, after distance learning is completed, a kmeans algorithm is used, similarity information is not considered, vehicle track samples are clustered, a main motion mode of vehicles on a road section is obtained, and the vehicle track with the measured distance larger than a threshold value is considered to be abnormal.
4. SSIM-based dual-mode fusion:
when a vehicle encounters an abnormal event, it typically decelerates to a full standstill. Therefore, a combination of the detection results of the static mode and the dynamic mode is required. And SSIM is introduced to judge whether the vehicles analyzed twice are the same vehicle. The output of the last fully connected layer of the two classifiers is used as a feature. And comparing the cosine similarity of the characteristic values of the static mode abnormal vehicle and the motion mode abnormal vehicle, and if the cosine similarity is higher than a threshold value, determining that the vehicles are the same vehicle. SSIM (structural similarity) calculation procedure is as follows:
measuring the similarity of the images from three aspects of brightness, contrast and structure respectively, wherein
、
Respectively representing images
And
the average value of (a) of (b),
，
respectively representing images
And
the variance of (a) is determined,
representing images
And
the covariance of (a), i.e.:
for equations (14) (15) (16),
、
、
is constant, usually taken
Generally, take
，
，
Then, then
The SSIM value range [0,1], wherein the larger the value is, the higher the similarity is. After the vehicles compare the similarity of the SSIM, the identification results of the static mode abnormality detection method and the motion mode abnormality detection method are fused, and the confidence coefficient and the occurrence time of an abnormal event are obtained:
wherein,
，
refers to the confidence level of a static vehicle and a moving vehicle,
is the final score of the entire system.
，
Means twoIn the method, the occurrence time of the abnormal event,
is the finally determined abnormality occurrence time.
Representing the weight of the static vehicle detection results.
The invention also provides a traffic anomaly detection system based on the dual-mode, which comprises the following steps: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the bimodal-based traffic anomaly detection method as described above when executing the computer program.
The processor and memory may be connected by a bus or other means.
The memory, which is a non-transitory computer readable storage medium, may be used to store non-transitory software programs as well as non-transitory computer executable programs. Further, the memory may include high speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory optionally includes memory located remotely from the processor, and these remote memories may be connected to the processor through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
It should be noted that the bimodal-based traffic anomaly detection system in this embodiment may include a service processing module, an edge database, a server version information register, and a data synchronization module, and when the processor executes a computer program, the bimodal-based traffic anomaly detection method applied to the bimodal-based traffic anomaly detection system is implemented.
The above-described embodiments of the apparatus are merely illustrative, wherein the units illustrated as separate components may or may not be physically separate, i.e. may be located in one place, or may also be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
Furthermore, an embodiment of the present invention further provides a computer-readable storage medium, which stores computer-executable instructions, which are executed by a processor or a controller, for example, by a processor in the above-mentioned terminal embodiment, and can make the processor execute the traffic anomaly detection method based on the dual mode in the above-mentioned embodiment.
One of ordinary skill in the art will appreciate that all or some of the steps, systems, and methods disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as is well known to those of ordinary skill in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can accessed by a computer. In addition, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media as known to those skilled in the art.
While the preferred embodiments of the present invention have been described in detail, it will be understood by those skilled in the art that the foregoing and various other changes, omissions and deviations in the form and detail thereof may be made without departing from the scope of this invention.
The above-described embodiments of the present invention should not be construed as limiting the scope of the present invention. Any other corresponding changes and modifications made according to the technical idea of the present invention should be included in the protection scope of the claims of the present invention.
Claims (10)
1. A traffic abnormity detection method based on dual modes is characterized by comprising the following steps:
acquiring traffic video stream data, wherein the video stream data comprises time data and picture frames;
detecting and identifying static vehicles in picture frames of the traffic video stream data;
detecting dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking vehicles with abnormal tracks;
comparing the characteristics of the static vehicle and the abnormal track vehicle in the aspects of brightness, contrast and structure, calculating to obtain the similarity of the static vehicle and the abnormal track vehicle, and acquiring the confidence coefficient and the event occurrence time of an abnormal event according to the similarity and the time data.
2. The dual-mode based traffic anomaly detection method according to claim 1, wherein said detecting and identifying static vehicles in picture frames of said traffic video stream data comprises the steps of:
calculating the weighted sum of input frames in the traffic video stream data, enhancing the static part of a picture, inhibiting the moving part of the picture, and extracting all picture frames containing static vehicle data in the traffic video stream data;
detecting a static vehicle in the picture frame using a yolo algorithm.
3. The bimodal-based traffic anomaly detection method according to claim 2, wherein said weighted sum of input frames in traffic video stream data is calculated by the formula:
wherein:
an ith frame representing a video;
is the average from frame 1 to frame i;
is the weight value of the image frame.
4. The dual-mode-based traffic anomaly detection method according to claim 2, wherein the step of detecting the static vehicles in the picture frame by using the yolo algorithm comprises the following steps:
preprocessing the picture frame and setting the picture frame to be a preset size;
the information of the previous picture frame is reserved through a residual error unit, and the dimensionality of the tensor is spliced and expanded through a tensor mode;
extracting image features of the picture frame through a convolutional layer to detect the static vehicle in the picture frame.
5. The dual-mode-based traffic anomaly detection method according to claim 1, wherein the detection of the dynamic vehicles in the traffic video stream data, tracking and identifying the dynamic vehicles and marking abnormal track vehicles comprises the following steps:
segmenting dynamic vehicles in the traffic video stream data by using a semantic segmentation algorithm;
tracking the dynamic vehicle using a target tracking algorithm;
and calculating the track of the dynamic vehicle, carrying out cluster analysis on the track to obtain a main motion mode of the dynamic vehicle, and marking the dynamic vehicle deviating from the main motion mode as an abnormal track vehicle.
6. The dual-modality based traffic anomaly detection method of claim 5, wherein the semantic segmentation algorithm is used to segment out dynamic vehicles in the traffic video stream data, comprising the steps of:
preprocessing the picture frame;
extracting a characteristic diagram of the picture frame through a convolution layer of a mask-rcnn network;
inputting the feature map into an RPN network, and calculating a candidate region of the picture frame;
pooling the feature maps of the candidate regions to a fixed size by a ROIAlign operation;
and in a mask-rcndn classification regression network, classifying the feature graph and regressing a target frame, and generating a mask through full-connection layer detection to segment the dynamic vehicle.
7. The dual-modality based traffic anomaly detection method of claim 5, wherein the tracking of the dynamic vehicle using a target tracking algorithm includes the steps of:
predicting to obtain the vehicle state information of the current frame according to the speed and position information of the dynamic vehicle in the previous frame and based on a Kalman filtering algorithm;
estimating the real state of the current dynamic vehicle according to the detection result and the prediction result of the current frame;
extracting a feature vector of the dynamic vehicle in a target frame by using a feature extraction network;
calculating the cosine distance of the characteristic vectors of the front frame and the rear frame as a quantized value of the similarity, matching the vehicles in the detection frames of the front frame and the rear frame, and distributing the same serial number to the dynamic vehicles with the matching degree above a preset threshold value.
8. The bimodal-based traffic anomaly detection method according to claim 1, wherein the calculating of the similarity between the static vehicle and the anomalous trajectory vehicle comprises the steps of:
and comparing the cosine similarity of the characteristic values of the static vehicle and the abnormal track vehicle, and when the cosine similarity is higher than a threshold value, determining that the static vehicle and the abnormal track vehicle are the same vehicle.
9. A dual-modality based traffic anomaly detection system, comprising: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the bimodal-based traffic anomaly detection method according to any one of claims 1 to 8 when executing the computer program.
10. A computer-readable storage medium storing computer-executable instructions for causing a computer to perform the bimodal-based traffic anomaly detection method as recited in any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210029611.3A CN114049771A (en) | 2022-01-12 | 2022-01-12 | Bimodal-based traffic anomaly detection method and system and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210029611.3A CN114049771A (en) | 2022-01-12 | 2022-01-12 | Bimodal-based traffic anomaly detection method and system and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114049771A true CN114049771A (en) | 2022-02-15 |
Family
ID=80196245
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210029611.3A Pending CN114049771A (en) | 2022-01-12 | 2022-01-12 | Bimodal-based traffic anomaly detection method and system and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114049771A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023207742A1 (en) * | 2022-04-28 | 2023-11-02 | 南京理工大学 | Method and system for detecting anomalous traffic behavior |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102737385A (en) * | 2012-04-24 | 2012-10-17 | 中山大学 | Video target tracking method based on CAMSHIFT and Kalman filtering |
| CN103377555A (en) * | 2012-04-25 | 2013-10-30 | 施乐公司 | Method and system for automatically detecting anomalies at a traffic intersection |
| CN104881661A (en) * | 2015-06-23 | 2015-09-02 | 河北工业大学 | Vehicle detection method based on structure similarity |
| CN106296729A (en) * | 2016-07-27 | 2017-01-04 | 南京华图信息技术有限公司 | The REAL TIME INFRARED THERMAL IMAGE imaging ground moving object tracking of a kind of robust and system |
| CN106652445A (en) * | 2016-11-15 | 2017-05-10 | 成都通甲优博科技有限责任公司 | Road traffic accident judging method and device |
| CN109285341A (en) * | 2018-10-31 | 2019-01-29 | 中电科新型智慧城市研究院有限公司 | A kind of urban road vehicle exception stagnation of movement detection method based on real-time video |
| CN109934075A (en) * | 2017-12-19 | 2019-06-25 | 杭州海康威视数字技术股份有限公司 | Accident detection method, apparatus, system and electronic equipment |
| CN110415277A (en) * | 2019-07-24 | 2019-11-05 | 中国科学院自动化研究所 | Based on light stream and the multi-target tracking method of Kalman filtering, system, device |
| CN110569702A (en) * | 2019-02-14 | 2019-12-13 | 阿里巴巴集团控股有限公司 | Video stream processing method and device |
| CN111105437A (en) * | 2018-10-29 | 2020-05-05 | 西安宇视信息科技有限公司 | Vehicle track abnormity judgment method and device |
| CN111626277A (en) * | 2020-08-03 | 2020-09-04 | 杭州智诚惠通科技有限公司 | Vehicle tracking method and device based on over-station inter-modulation index analysis |
| CN112633228A (en) * | 2020-12-31 | 2021-04-09 | 北京市商汤科技开发有限公司 | Parking detection method, device, equipment and storage medium |
| CN112884742A (en) * | 2021-02-22 | 2021-06-01 | 山西讯龙科技有限公司 | Multi-algorithm fusion-based multi-target real-time detection, identification and tracking method |
| CN113409587A (en) * | 2021-06-16 | 2021-09-17 | 北京字跳网络技术有限公司 | Abnormal vehicle detection method, device, equipment and storage medium |
| CN113792586A (en) * | 2021-08-04 | 2021-12-14 | 武汉市公安局交通管理局 | Vehicle accident detection method and device and electronic equipment |
-
2022
- 2022-01-12 CN CN202210029611.3A patent/CN114049771A/en active Pending
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102737385A (en) * | 2012-04-24 | 2012-10-17 | 中山大学 | Video target tracking method based on CAMSHIFT and Kalman filtering |
| CN103377555A (en) * | 2012-04-25 | 2013-10-30 | 施乐公司 | Method and system for automatically detecting anomalies at a traffic intersection |
| CN104881661A (en) * | 2015-06-23 | 2015-09-02 | 河北工业大学 | Vehicle detection method based on structure similarity |
| CN106296729A (en) * | 2016-07-27 | 2017-01-04 | 南京华图信息技术有限公司 | The REAL TIME INFRARED THERMAL IMAGE imaging ground moving object tracking of a kind of robust and system |
| CN106652445A (en) * | 2016-11-15 | 2017-05-10 | 成都通甲优博科技有限责任公司 | Road traffic accident judging method and device |
| CN109934075A (en) * | 2017-12-19 | 2019-06-25 | 杭州海康威视数字技术股份有限公司 | Accident detection method, apparatus, system and electronic equipment |
| CN111105437A (en) * | 2018-10-29 | 2020-05-05 | 西安宇视信息科技有限公司 | Vehicle track abnormity judgment method and device |
| CN109285341A (en) * | 2018-10-31 | 2019-01-29 | 中电科新型智慧城市研究院有限公司 | A kind of urban road vehicle exception stagnation of movement detection method based on real-time video |
| CN110569702A (en) * | 2019-02-14 | 2019-12-13 | 阿里巴巴集团控股有限公司 | Video stream processing method and device |
| CN110415277A (en) * | 2019-07-24 | 2019-11-05 | 中国科学院自动化研究所 | Based on light stream and the multi-target tracking method of Kalman filtering, system, device |
| CN111626277A (en) * | 2020-08-03 | 2020-09-04 | 杭州智诚惠通科技有限公司 | Vehicle tracking method and device based on over-station inter-modulation index analysis |
| CN112633228A (en) * | 2020-12-31 | 2021-04-09 | 北京市商汤科技开发有限公司 | Parking detection method, device, equipment and storage medium |
| CN112884742A (en) * | 2021-02-22 | 2021-06-01 | 山西讯龙科技有限公司 | Multi-algorithm fusion-based multi-target real-time detection, identification and tracking method |
| CN113409587A (en) * | 2021-06-16 | 2021-09-17 | 北京字跳网络技术有限公司 | Abnormal vehicle detection method, device, equipment and storage medium |
| CN113792586A (en) * | 2021-08-04 | 2021-12-14 | 武汉市公安局交通管理局 | Vehicle accident detection method and device and electronic equipment |
Non-Patent Citations (4)
| Title |
|---|
| C. RAJESH BABU等: "Vehicle Traffic Analysis Using Yolo", 《EURASIAN JOURNAL OF ANALYTICAL CHEMISTRY》 * |
| SHAOFENG LIU等: "Fuzzy Control Reversing System Based on Visual Information", 《SPRINGER》 * |
| YAN XU等: "Dual-Mode Vehicle Motion Pattern Learning for High Performance Road Traffic Anomaly Detection", 《IEEE》 * |
| 董洪义: "《深度学习之PyTorch物体检测实战》", 31 March 2020 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023207742A1 (en) * | 2022-04-28 | 2023-11-02 | 南京理工大学 | Method and system for detecting anomalous traffic behavior |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10212397B2 (en) | Abandoned object detection apparatus and method and system | |
| JP5815910B2 (en) | Methods, systems, products, and computer programs for multi-queue object detection and analysis (multi-queue object detection and analysis) | |
| Ling et al. | Identifying parking spaces & detecting occupancy using vision-based IoT devices | |
| Barcellos et al. | A novel video based system for detecting and counting vehicles at user-defined virtual loops | |
| JP5127392B2 (en) | Classification boundary determination method and classification boundary determination apparatus | |
| US20140376769A1 (en) | Method for detecting large size and passenger vehicles from fixed cameras | |
| CN109344886B (en) | Occlusion number plate distinguishing method based on convolutional neural network | |
| KR20120032897A (en) | Method and system for detecting object in input image | |
| CN112633255B (en) | Target detection method, device and equipment | |
| CN112435276B (en) | Vehicle tracking method and device, intelligent terminal and storage medium | |
| CN110991447B (en) | Train number accurate positioning and identifying method based on deep learning | |
| CN113515968A (en) | Method, device, equipment and medium for detecting street abnormal event | |
| CN111079621A (en) | Method and device for detecting object, electronic equipment and storage medium | |
| Saha et al. | Developing a framework for vehicle detection, tracking and classification in traffic video surveillance | |
| CN114937248A (en) | Vehicle tracking method and device for cross-camera, electronic equipment and storage medium | |
| CN114049771A (en) | Bimodal-based traffic anomaly detection method and system and storage medium | |
| US20200394802A1 (en) | Real-time object detection method for multiple camera images using frame segmentation and intelligent detection pool | |
| CN114141022A (en) | Emergency lane occupation behavior detection method and device, electronic equipment and storage medium | |
| WO2017077261A1 (en) | A monocular camera cognitive imaging system for a vehicle | |
| CN112906495B (en) | Target detection method and device, electronic equipment and storage medium | |
| Zaman et al. | Deep Learning Approaches for Vehicle and Pedestrian Detection in Adverse Weather | |
| CN110781710B (en) | Target object clustering method and device | |
| Maaloul | Video-based algorithms for accident detections | |
| CN112686136A (en) | Object detection method, device and system | |
| CN112906424B (en) | Image recognition method, device and equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220215 |
|
| RJ01 | Rejection of invention patent application after publication |