CN111160203A - Loitering and lingering behavior analysis method based on head and shoulder model and IOU tracking - Google Patents
Loitering and lingering behavior analysis method based on head and shoulder model and IOU tracking Download PDFInfo
- Publication number
- CN111160203A CN111160203A CN201911342538.XA CN201911342538A CN111160203A CN 111160203 A CN111160203 A CN 111160203A CN 201911342538 A CN201911342538 A CN 201911342538A CN 111160203 A CN111160203 A CN 111160203A
- Authority
- CN
- China
- Prior art keywords
- head
- shoulder
- target
- pedestrian
- tracking
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
Abstract
A loitering lingering behavior analysis method based on a head and shoulder model and IOU tracking comprises the following steps: s1: the pedestrian head and shoulder data set is established by crawling a pedestrian picture on the network, intercepting the picture containing the pedestrian in the monitoring video, marking and cleaning the head and shoulder of the pedestrian in the picture; s2: detecting the monitoring video, and returning to the head and shoulders of the pedestrian detected in the video picture; s4: carrying out multi-target tracking on the head and the shoulder of the pedestrian by adopting an Iou-net tracking algorithm; s5: tracking the motion track path of the target, and comparing the motion track path with a set threshold; s6: outputting a loitering target. In the invention, a head and shoulder data set is established, and a deep learning detection algorithm is utilized for training. The pedestrian head and shoulder in the monitoring video are detected through the trained model, the detected head and shoulder are subjected to multi-target tracking by combining an Iou-net tracking algorithm, and whether the target lingers or not is analyzed by judging the motion track of the target.
Description
Technical Field
The invention relates to the field of urban monitoring video analysis in smart city construction, in particular to a loitering and sojourn behavior analysis method based on a head-shoulder model and IOU tracking.
Background
The existing main method for analyzing the behavior of the pedestrian loitering lingering is as follows:
the traditional background difference method is used for detecting a moving target and analyzing the track to determine whether a lingering behavior exists. And performing secondary matching according to the target position and the histogram, and judging whether the target has a lingering behavior or not by judging the times and time of the target appearing in the monitoring video. And (3) judging whether the loitering behavior exists or not by analyzing the track by using a pedestrian detection and tracking combined method.
The main disadvantages of the existing analysis of the behavior of the pedestrian loitering lingering are as follows:
1. the traditional background difference method needs to store a large amount of target data in the modeling process, and is easy to cause the defects in the aspects of background updating, disturbance, shadow suppression and the like.
2. The method for performing secondary matching by using the target position and the histogram is easy to cause inaccurate matching due to the influence of background disturbance, thereby causing low judgment precision.
3. By using the pedestrian detection method, the detection rate is easily low and the target tracking is lost under the condition that the pedestrian is shielded, so that the detection effect is poor.
Disclosure of Invention
Objects of the invention
In order to solve the technical problems in the background art, the invention provides a loitering lingering behavior analysis method based on a head-shoulder model and IOU tracking. The pedestrian head and shoulder in the monitoring video are detected through the trained model, the detected head and shoulder are subjected to multi-target tracking by combining an Iou-net tracking algorithm, and whether the target lingers or not is analyzed by judging the motion track of the target.
(II) technical scheme
In order to solve the problems, the invention provides a loitering and sojousting behavior analysis method based on a head-shoulder model and IOU tracking, which comprises the following steps:
s1: the pedestrian head and shoulder data set is established by crawling a pedestrian picture on the network, intercepting the picture containing the pedestrian in the monitoring video, marking and cleaning the head and shoulder of the pedestrian in the picture;
s2: training and parameter adjusting are carried out on the established head and shoulder data set by adopting a deep learning algorithm YoIo-v3 to obtain a head and shoulder model;
s3: detecting the monitoring video by adopting a YoIo-v3 algorithm and combining a head-shoulder model, and returning the detected head and shoulder of the pedestrian in the video picture;
s4: carrying out multi-target tracking on the head and the shoulder of the pedestrian by adopting an Iou-net tracking algorithm, and carrying out track processing and calculation on the tracked target;
s5: drawing a motion track path of a tracking target in video monitoring, and comparing a motion track calculation result with a set threshold value;
if the motion track of the tracked target is larger than a set threshold value, the target is considered to have a loitering behavior in the current picture, the system calculates the position information of the target, and marks the target in the video picture;
if the motion track of the tracked target is smaller than a set threshold value, the target is considered to have no loitering behavior in the current picture, and no processing is performed;
s6: outputting loitering targets
Preferably, in S1, 14000 pictures are contained in total.
In the invention, the head and shoulder model is trained by adopting the self-built head and shoulder data set, and the detection is carried out by utilizing the deep learning algorithm YoIo-v3, so that the method has the characteristics of high speed and high precision. Compared with the traditional pedestrian detection algorithm, the method adopts the head and shoulder detection, so that the problem of reduced detection precision under the condition of serious pedestrian shielding can be solved.
According to the invention, Iou-net is adopted for multi-target tracking, so that the problems of low detection rate, easy loss of moving targets and the like caused by traditional methods such as motion background analysis and the like are solved, and the tracking precision and the tracking speed are greatly improved.
According to the pedestrian detection method, the accuracy of pedestrian detection, the tracking speed and the accuracy of loitering behavior judgment are improved by combining head and shoulder detection, Iou-net tracking and motion track path analysis. The pedestrian loitering behavior in the monitoring video can be analyzed in real time and efficiently.
Drawings
Fig. 1 is a schematic flow chart of a wandering and staying behavior analysis method based on a head-shoulder model and IOU tracking according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings in conjunction with the following detailed description. It should be understood that the description is intended to be exemplary only, and is not intended to limit the scope of the present invention. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present invention.
As shown in fig. 1, the loitering lingering behavior analysis method based on the head-shoulder model and the IOU tracking provided by the invention comprises the following steps:
s1: the pedestrian head and shoulder data set is established by crawling a pedestrian picture on the network, intercepting the picture containing the pedestrian in the monitoring video, marking and cleaning the head and shoulder of the pedestrian in the picture;
s2: training and parameter adjusting are carried out on the established head and shoulder data set by adopting a deep learning algorithm YoIo-v3 to obtain a head and shoulder model;
s3: detecting the monitoring video by adopting a YoIo-v3 algorithm and combining a head-shoulder model, and returning the detected head and shoulder of the pedestrian in the video picture;
s4: carrying out multi-target tracking on the head and the shoulder of the pedestrian by adopting an Iou-net tracking algorithm, and carrying out track processing and calculation on the tracked target;
s5: drawing a motion track path of a tracking target in video monitoring, and comparing a motion track calculation result with a set threshold value;
if the motion track of the tracked target is larger than a set threshold value, the target is considered to have a loitering behavior in the current picture, the system calculates the position information of the target, and marks the target in the video picture;
if the motion track of the tracked target is smaller than a set threshold value, the target is considered to have no loitering behavior in the current picture, and no processing is performed;
s6: outputting a loitering target.
In an alternative embodiment, a total of 14000 pictures are available in S1.
In the invention, the head and shoulder model is trained by adopting the self-built head and shoulder data set, and the detection is carried out by utilizing the deep learning algorithm YoIo-v3, so that the method has the characteristics of high speed and high precision. Compared with the traditional pedestrian detection algorithm, the method adopts the head and shoulder detection, so that the problem of reduced detection precision under the condition of serious pedestrian shielding can be solved.
According to the invention, Iou-net is adopted for multi-target tracking, so that the problems of low detection rate, easy loss of moving targets and the like caused by traditional methods such as motion background analysis and the like are solved, and the tracking precision and the tracking speed are greatly improved.
According to the pedestrian detection method, the accuracy of pedestrian detection, the tracking speed and the accuracy of loitering behavior judgment are improved by combining head and shoulder detection, Iou-net tracking and motion track path analysis. The pedestrian loitering behavior in the monitoring video can be analyzed in real time and efficiently.
In an optional embodiment, the extracted image feature points are subjected to dynamic planning trajectory tracking, that is, optimization processing is performed on S1-S3, specifically: the head and shoulder targets under the current monitoring picture are not detected, track tracking is directly carried out on candidate target pixels in the image, and evaluation analysis is carried out on each track to judge whether the loitering lingering behavior exists.
It should be noted that in the loitering lingering behavior analysis based on the head and shoulder model and the IOU tracking, the loitering lingering behavior analysis needs to rely on comprehensive analysis such as head and shoulder detection, target tracking, motion trajectory path analysis and the like to determine whether loitering lingering lingers, and by adopting the method, the flow of detecting the loitering lingering behavior can be simplified, but the extracted feature points are possibly inaccurate due to the picture quality problem, the illumination problem and the like of the monitoring video, and the final detection result is inaccurate.
It is to be understood that the above-described embodiments of the present invention are merely illustrative of or explaining the principles of the invention and are not to be construed as limiting the invention. Therefore, any modification, equivalent replacement, improvement and the like made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention. Further, it is intended that the appended claims cover all such variations and modifications as fall within the scope and boundaries of the appended claims or the equivalents of such scope and boundaries.
Claims (2)
1. A loitering lingering behavior analysis method based on a head and shoulder model and IOU tracking is characterized by comprising the following steps:
s1: the pedestrian head and shoulder data set is established by crawling a pedestrian picture on the network, intercepting the picture containing the pedestrian in the monitoring video, marking and cleaning the head and shoulder of the pedestrian in the picture;
s2: training and parameter adjusting are carried out on the established head and shoulder data set by adopting a deep learning algorithm YoIo-v3 to obtain a head and shoulder model;
s3: detecting the monitoring video by adopting a YoIo-v3 algorithm and combining a head-shoulder model, and returning the detected head and shoulder of the pedestrian in the video picture;
s4: carrying out multi-target tracking on the head and the shoulder of the pedestrian by adopting an Iou-net tracking algorithm, and carrying out track processing and calculation on the tracked target;
s5: drawing a motion track path of a tracking target in video monitoring, and comparing a motion track calculation result with a set threshold value;
if the motion track of the tracked target is larger than a set threshold value, the target is considered to have a loitering behavior in the current picture, the system calculates the position information of the target, and marks the target in the video picture;
if the motion track of the tracked target is smaller than a set threshold value, the target is considered to have no loitering behavior in the current picture, and no processing is performed;
s6: outputting a loitering target.
2. The method for analyzing wandering behavior based on a head-shoulder model and IOU tracking according to claim 1, wherein in S1, 14000 pictures are provided in total.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911342538.XA CN111160203B (en) | 2019-12-23 | 2019-12-23 | Loitering and stay behavior analysis method based on head-shoulder model and IOU tracking |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911342538.XA CN111160203B (en) | 2019-12-23 | 2019-12-23 | Loitering and stay behavior analysis method based on head-shoulder model and IOU tracking |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111160203A true CN111160203A (en) | 2020-05-15 |
| CN111160203B CN111160203B (en) | 2023-05-16 |
Family
ID=70558195
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911342538.XA Active CN111160203B (en) | 2019-12-23 | 2019-12-23 | Loitering and stay behavior analysis method based on head-shoulder model and IOU tracking |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111160203B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111860318A (en) * | 2020-07-20 | 2020-10-30 | 杭州品茗安控信息技术股份有限公司 | Construction site pedestrian loitering detection method, device, equipment and storage medium |
| CN112232211A (en) * | 2020-10-16 | 2021-01-15 | 山东科技大学 | Intelligent video monitoring system based on deep learning |
| CN112329671A (en) * | 2020-11-11 | 2021-02-05 | 润联软件系统(深圳)有限公司 | Pedestrian running behavior detection method based on deep learning and related components |
| CN112541432A (en) * | 2020-12-11 | 2021-03-23 | 上海品览数据科技有限公司 | Video livestock identity authentication system and method based on deep learning |
| CN112633205A (en) * | 2020-12-28 | 2021-04-09 | 北京眼神智能科技有限公司 | Pedestrian tracking method and device based on head and shoulder detection, electronic equipment and storage medium |
| CN113762231A (en) * | 2021-11-10 | 2021-12-07 | 中电科新型智慧城市研究院有限公司 | End-to-end multi-pedestrian posture tracking method and device and electronic equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090017886A1 (en) * | 2007-07-11 | 2009-01-15 | Dr. Elliot McGucken | System and method for creating exalted video games and virtual realities wherein ideas have consequences |
| WO2015131734A1 (en) * | 2014-07-25 | 2015-09-11 | 中兴通讯股份有限公司 | Method, device, and storage medium for pedestrian counting in forward looking surveillance scenario |
| CN109117721A (en) * | 2018-07-06 | 2019-01-01 | 江西洪都航空工业集团有限责任公司 | A kind of pedestrian hovers detection method |
-
2019
- 2019-12-23 CN CN201911342538.XA patent/CN111160203B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090017886A1 (en) * | 2007-07-11 | 2009-01-15 | Dr. Elliot McGucken | System and method for creating exalted video games and virtual realities wherein ideas have consequences |
| WO2015131734A1 (en) * | 2014-07-25 | 2015-09-11 | 中兴通讯股份有限公司 | Method, device, and storage medium for pedestrian counting in forward looking surveillance scenario |
| CN109117721A (en) * | 2018-07-06 | 2019-01-01 | 江西洪都航空工业集团有限责任公司 | A kind of pedestrian hovers detection method |
Non-Patent Citations (1)
| Title |
|---|
| 朱梦哲;冯瑞;: "基于三维模型的行人徘徊行为检测算法" * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111860318A (en) * | 2020-07-20 | 2020-10-30 | 杭州品茗安控信息技术股份有限公司 | Construction site pedestrian loitering detection method, device, equipment and storage medium |
| CN112232211A (en) * | 2020-10-16 | 2021-01-15 | 山东科技大学 | Intelligent video monitoring system based on deep learning |
| CN112329671A (en) * | 2020-11-11 | 2021-02-05 | 润联软件系统(深圳)有限公司 | Pedestrian running behavior detection method based on deep learning and related components |
| CN112541432A (en) * | 2020-12-11 | 2021-03-23 | 上海品览数据科技有限公司 | Video livestock identity authentication system and method based on deep learning |
| CN112633205A (en) * | 2020-12-28 | 2021-04-09 | 北京眼神智能科技有限公司 | Pedestrian tracking method and device based on head and shoulder detection, electronic equipment and storage medium |
| CN113762231A (en) * | 2021-11-10 | 2021-12-07 | 中电科新型智慧城市研究院有限公司 | End-to-end multi-pedestrian posture tracking method and device and electronic equipment |
| CN113762231B (en) * | 2021-11-10 | 2022-03-22 | 中电科新型智慧城市研究院有限公司 | End-to-end multi-pedestrian posture tracking method and device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111160203B (en) | 2023-05-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111160203B (en) | Loitering and stay behavior analysis method based on head-shoulder model and IOU tracking | |
| US10769480B2 (en) | Object detection method and system | |
| CN108986064B (en) | People flow statistical method, equipment and system | |
| CN104134222B (en) | Traffic flow monitoring image detecting and tracking system and method based on multi-feature fusion | |
| CN102542289B (en) | Pedestrian volume statistical method based on plurality of Gaussian counting models | |
| CN106778712B (en) | Multi-target detection and tracking method | |
| CN109145708B (en) | Pedestrian flow statistical method based on RGB and D information fusion | |
| CN104615986B (en) | The method that pedestrian detection is carried out to the video image of scene changes using multi-detector | |
| CN103646257B (en) | A kind of pedestrian detection and method of counting based on video monitoring image | |
| RU2013143669A (en) | METHOD AND SYSTEM FOR USING PORTRAITS WHEN MONITORING MOVING OBJECTS BY VIDEO RECORDING | |
| CN104318263A (en) | Real-time high-precision people stream counting method | |
| CN105760846A (en) | Object detection and location method and system based on depth data | |
| CN104835147A (en) | Method for detecting crowded people flow in real time based on three-dimensional depth map data | |
| CN109711256B (en) | Low-altitude complex background unmanned aerial vehicle target detection method | |
| CN108596157B (en) | Crowd disturbance scene detection method and system based on motion detection | |
| WO2020258978A1 (en) | Object detection method and device | |
| CN110222579B (en) | Video object counting method combining motion law and target detection | |
| CN106504274A (en) | A kind of visual tracking method and system based under infrared camera | |
| CN111968159A (en) | Simple and universal fish video image track tracking method | |
| CN108830204B (en) | Method for detecting abnormality in target-oriented surveillance video | |
| CN111640135A (en) | TOF camera pedestrian counting method based on hardware front end | |
| KR102434397B1 (en) | Real time multi-object tracking device and method by using global motion | |
| Li et al. | Object-oriented anomaly detection in surveillance videos | |
| CN116153086A (en) | Multi-path traffic accident and congestion detection method and system based on deep learning | |
| CN108765463A (en) | A kind of moving target detecting method calmodulin binding domain CaM extraction and improve textural characteristics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |