CN108986158A - A kind of across the scene method for tracing identified again based on target and device and Computer Vision Platform - Google Patents
A kind of across the scene method for tracing identified again based on target and device and Computer Vision Platform Download PDFInfo
- Publication number
- CN108986158A CN108986158A CN201810937026.7A CN201810937026A CN108986158A CN 108986158 A CN108986158 A CN 108986158A CN 201810937026 A CN201810937026 A CN 201810937026A CN 108986158 A CN108986158 A CN 108986158A
- Authority
- CN
- China
- Prior art keywords
- target
- track
- scene
- camera
- module
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
- G06T7/55—Depth or shape recovery from multiple images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30232—Surveillance
Abstract
The present invention relates to a kind of across scene method for tracing identified again based on target and device and Computer Vision Platform;The wherein video data of n camera this method comprises: acquisition in real time is mutually related, obtains the road n video data;The target occurred in the video data of the road n is detected, the target occurred in each scene is captured;The target occurred in each scene of the capture is tracked in subsequent frames, obtains track of the target of the appearance in each camera;Known again by target otherwise, the track that same target is belonged in n camera is together in series, cluster obtains target trajectory;According to the target trajectory that cluster obtains, across the scene track of target is determined;Electronic map is made, across the scene track of the target of the determination is shown on the electronic map.The present invention is in the monitoring scene with association multi-cam, by to target each single camera capture after, target is again identified that using the hierarchical clustering of track, same target is connected into cross-scenario track in the track of single scene, realizes the tracking to target under across scene.
Description
Technical field
The present invention relates to technical field of video monitoring more particularly to a kind of across scene method for tracing identified again based on target
With device and Computer Vision Platform.
Background technique
In building smart city, under the development trend of safe city, the figure of intelligent monitor system appears in more and more
Field.Traditional telesecurity monitoring system is often based upon single camera or camera array, not smart enoughization, certain
The occasion of monitoring objective behavior act rule can pass through machine learning, the modern technologies such as automatic detection and tracking of target completely
To reduce the consumption of manpower.
Computer vision field, on the basis of studying advanced visual target tracking algorithm, to the long-term follow of single goal,
The problems such as target tracker, target detection track device, in conjunction with new academic frontier knowledge, it is proposed that innovation intelligent algorithm is finally applied to
Trans-regional human body video frequency object tracking intelligent monitor system carries out single goal long-term follow, positioning and identification, has preferable skill
Art performance.
The long-term follow of single goal refers to after single tracked target has been determined using a bounding box, in continuous frame
The interior process that detection recognition and tracking is carried out to the target.During tracking, tracking need to overcome environment and by
The continually changing interference for tracking target, realizes long-term target following.The interference for generally requiring solution includes ambient lighting
Variation, noise, the movement of video sensor, occlusion issue, target leaves scene and target is again introduced into scene.Target is left
Scene and target are again introduced into the detection and lasting tracking of scene.For problem, to the mesh in single camera covering scope
Mark detection and tracking, are calculated whether by histograms of oriented gradients, initial characteristics point, Particle Swarm, the nearest, light stream in space etc.
Method all it has been proposed that.But it is continuously tracked and predicts that algorithm above just can not using multiple cameras progress target when desired
Purpose is directly reached, due to being significantly increased for data volume, less with wanting to carry out real-time tracking.
It is based on the track for recording target under single camera lens, without very mostly at present to the target following under monitoring scene
The resource that good utilization entirely monitors.It is complicated and changeable due to different scenes for cross-scenario situation, it does not solve such
The solution that target tracks for a long time.
Summary of the invention
In view of this, social public security is safeguarded effectively to use each camera resource under monitoring scene,
The present invention provides a kind of across scene method for tracing and device identified again based on target, same to occurring in associated more scenes
Target is effectively gone here and there the track of the target according to time-space relationship by knowing again otherwise after the capture of each camera
Connection, and shown in electronic map, it realizes and cross-scenario tracking is carried out to target.
To solve the above problems, technical solution provided by the invention is as follows:
A kind of across scene method for tracing identified again based on target characterized by comprising
The video data for acquiring the n camera that be mutually related in real time, obtains the road n video data;
The target occurred in the video data of the road n is detected, the target occurred in each scene is captured;
The target occurred in each scene of the capture is tracked in subsequent frames, obtains the target of the appearance
Track in each camera;
Known again by target otherwise, the track that same target is belonged in n camera is together in series, cluster obtains
Target trajectory;
According to the target trajectory that cluster obtains, across the scene track of target is determined;
Electronic map is made, across the scene track of the target of the determination is shown on the electronic map.
Preferably, described the target occurred in the video data of the road n is detected, it captures and occurs in each scene
Target specifically includes:
By the way of Machine learning classifiers or deep learning, the target occurred in the video data of the road n is carried out
Detection captures all kinds of targets in each scene by extracting region candidate and classification.
Preferably, the target occurred in each scene to the capture tracks in subsequent frames, obtains institute
Track of the target of appearance in each camera is stated, is specifically included:
Using twin network and region candidate network struction monotrack device, then established respectively for each target respectively
Monotrack device, and then multiple target tracking is extended to, to obtain rail of the target of the appearance in each camera
Mark.
Preferably, described to be known again by target otherwise, the track that same target is belonged in n camera is connected
To come, cluster obtains target trajectory, it specifically includes:
Critical point detection is carried out to target, four, accurate upper and lower, left and right of target boundary is determined, obtains target accounting most
Big region of interest ROI;
Clarification of objective is extracted by way of deep learning;
The track for belonging to same target in n camera is together in series, cluster obtains target trajectory.
Preferably, described that clarification of objective is extracted by way of deep learning, it specifically includes:
Clarification of objective is extracted from by the way of encoding using unsupervised learning;
Alternatively,
Clarification of objective is extracted using convolutional neural networks.
Preferably, described that the track for belonging to same target in n camera is together in series, cluster obtains target trajectory,
It specifically includes:
According to the method for hierarchical clustering, by the track of each target see in mapping a bit, between track and track away from
From side is constituted, mode is cut by the figure of max-flow min-cut, all tracks for belonging to same target are got together, target is obtained
Track.
Preferably, the target trajectory obtained according to cluster, determines across the scene track of target, specifically includes:
According to the target trajectory that cluster obtains, the time sequencing occurred in conjunction with target in each track and spatial relation,
Determine the cross-scenario track of target.
Preferably, the production electronic map shows across the scene track of the target of the determination in the electronic map
On, it specifically includes:
Region where the described n camera that be mutually related makes electronic map, will on the electronic map
Across the scene track of target is mapped according to specific location, to show target in the running route of whole region.
The present invention also provides a kind of across scene follow-up mechanisms identified again based on target characterized by comprising
Video acquisition module obtains the road n video counts for acquiring the video data of the n camera that be mutually related in real time
According to;
Module of target detection captures in each scene for detecting to the target occurred in the video data of the road n
The target of appearance;
Target tracking module, the target for occurring in each scene to the capture track in subsequent frames,
Obtain track of the target of the appearance in each camera;
Target identification module again will belong to the rail of same target for knowing again otherwise by target in n camera
Mark is together in series, and cluster obtains target trajectory;
Track determining module, the target trajectory for being obtained according to cluster, determines across the scene track of target;
Track display module shows across the scene track of the target of the determination in the electricity for making electronic map
On sub- map.
Preferably, the module of target detection specifically includes:
Target detection submodule, for by the way of Machine learning classifiers or deep learning, to the road n video counts
It is detected according to the target of middle appearance;
Target acquistion submodule, for being caught by extracting region candidate and classification to all kinds of targets in each scene
It obtains.
Preferably, the target tracking module specifically includes:
Tracker constructs module, is each for utilizing twin network and region candidate network struction monotrack device
Target establishes respective monotrack device respectively;
Track generation module obtains the target of the appearance in each camera for extending to multiple target tracking
Track.
Preferably, identification module specifically includes the target again:
Target critical point detection module determines the accurate upper and lower, left and right of target for carrying out critical point detection to target
Four boundaries obtain the maximum region of interest ROI of target accounting;
Target's feature-extraction module, for extracting clarification of objective by way of deep learning;
Trajectory clustering module, for the track for belonging to same target in n camera to be together in series, cluster obtains target
Track.
Preferably, the target's feature-extraction module specifically includes:
First object characteristic extracting module, for extracting clarification of objective from by the way of encoding using unsupervised learning;
Alternatively,
Second target's feature-extraction module, for extracting clarification of objective using convolutional neural networks.
Preferably, the trajectory clustering module specifically includes:
A bit in mapping, rail are seen for the method according to hierarchical clustering in the track of each target by hierarchical clustering module
The distance between mark and track constitute side, cut mode by the figure of max-flow min-cut, will belong to all tracks of same target
It gets together, obtains target trajectory.
Preferably, the track determining module specifically includes:
Track association module, the target trajectory for being obtained according to cluster, the time occurred in conjunction with target in each track are suitable
Sequence and spatial relation determine the cross-scenario track of target.
Preferably, the track display module specifically includes:
Trajectory map module makes electronic map for the region where the n camera that be mutually related according to,
On the electronic map, across the scene track of the target is mapped according to specific location, to show target entire
The running route in region.
The present invention also provides a kind of Computer Vision Platforms, comprising: n camera and such as above-mentioned any one are based on mesh
Mark across the scene follow-up mechanism identified again.
The present invention proposes a kind of across the scene method for tracing identified again based on target and device and Computer Vision Platform, can
By across the scene comparison to target, the track of same target in different scenes to be brought together, realize to target across field
The tracking of scape.The advantage is that:
1, the detection method of all target exploitation deep learnings occurred in scene is extracted between each camera,
It can guarantee different scale, different angle, the target of different the ratio of width to height is disposable simultaneously to be extracted.
2, using the method for tracking target of twin network and region candidate network, guarantee the accurate tracking of target.
3, clarification of objective is extracted by deep learning autoencoder network, is identified again for target, enhances the Shandong of feature
Stick and ability to express.
4, it proposes to cluster across scene objects tracks by the way of hierarchical clustering, automatic for realizing target trajectory
Match, to realize across the scene tracking to target.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention without any creative labor, may be used also for those of ordinary skill in the art
To obtain other drawings based on these drawings.
Fig. 1 is a kind of a kind of embodiment across scene method for tracing identified again based on target of the embodiment of the present invention
Flow chart;
Fig. 2 is a kind of a kind of embodiment party for identifying cluster again based on target and obtaining target trajectory method of the embodiment of the present invention
The flow chart of formula;
Fig. 3 is a kind of a kind of embodiment across scene follow-up mechanism identified again based on target of the embodiment of the present invention;
Fig. 4 is a kind of a kind of embodiment of Computer Vision Platform of the embodiment of the present invention.
Specific embodiment
Basic thought of the invention is in the monitoring scene with association multi-cam, by singly taking the photograph to target each
After head capture, target is again identified that using the hierarchical clustering of track, same target is connected in the track of single scene
It is connected into cross-scenario track, realizes the tracking to target under across scene.
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description;Obviously, described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Referring to Fig. 1, Fig. 1 is a kind of across scene method for tracing identified again based on target provided in an embodiment of the present invention.Institute
State across the scene method for tracing identified again based on target, comprising:
101, the video data for acquiring the n camera that be mutually related in real time, obtains the road n video data.
102, the target occurred in the video data of the road n is detected, captures the target occurred in each scene.
In this step, by the way of Machine learning classifiers or deep learning, to occurring in the video data of the road n
Target detected, by extract region candidate and classification all kinds of targets in each scene are captured.
103, the target occurred in each scene of the capture is tracked in subsequent frames, obtains the appearance
Track of the target in each camera.
It in this step, is then each target using twin network and region candidate network struction monotrack device
Respective monotrack device is established respectively, and then extends to multiple target tracking, to obtain the target of the appearance each
Track in camera.
104, known again by target otherwise, the track that same target is belonged in n camera is together in series, clustered
Obtain target trajectory.
105, the target trajectory obtained according to cluster, determines across the scene track of target.
In this step, the target trajectory obtained according to cluster, the time sequencing occurred in conjunction with target in each track and sky
Between positional relationship, determine the cross-scenario track of target.
106, electronic map is made, across the scene track of the target of the determination is shown on the electronic map.
In this step, electronic map is made according to the region where the described n camera that be mutually related, in the electricity
On sub- map, across the scene track of the target is mapped according to specific location, to show target in whole region
Running route.
This method have association multi-cam monitoring scene in, by target each single camera capture after, benefit
Target is again identified that with the hierarchical clustering of track, same target is connected into cross-scenario rail in the track of single scene
Mark realizes the tracking to target under across scene.
Referring to fig. 2, Fig. 2 is that a kind of target that is based on provided in an embodiment of the present invention identifies that cluster obtains target trajectory method again
A kind of embodiment.It is described to be known again by target otherwise, the track that same target is belonged in n camera is connected
To come, cluster obtains target trajectory, it specifically includes:
201, critical point detection is carried out to target, determines four, accurate upper and lower, left and right of target boundary, obtains target and account for
Than maximum region of interest ROI;
202, clarification of objective is extracted by way of deep learning;
In this step, clarification of objective is extracted from by the way of encoding using unsupervised learning;
Alternatively, extracting clarification of objective using convolutional neural networks.
203, the track for belonging to same target in n camera is together in series, cluster obtains target trajectory.
In this step, according to the method for hierarchical clustering, a bit in mapping, track and rail are seen into the track of each target
The distance between mark constitutes side, cuts mode by the figure of max-flow min-cut, all tracks for belonging to same target are gathered one
It rises, obtains target trajectory.
This method extracts clarification of objective by deep learning mode, identifies again for target, enhances the robust of feature
Property and ability to express;Across scene objects tracks are clustered by the way of hierarchical clustering, realize the automatic of target trajectory
Matching, to realize across the scene tracking to target.
Referring to Fig. 3, Fig. 3 is a kind of showing across scene follow-up mechanism identified again based on target provided in an embodiment of the present invention
It is intended to;As shown in figure 3, a kind of across scene follow-up mechanism identified again based on target provided in an embodiment of the present invention, comprising:
Video acquisition module obtains the road n video counts for acquiring the video data of the n camera that be mutually related in real time
According to;
Module of target detection captures in each scene for detecting to the target occurred in the video data of the road n
The target of appearance;
Target tracking module, the target for occurring in each scene to the capture track in subsequent frames,
Obtain track of the target of the appearance in each camera;
Target identification module again will belong to the rail of same target for knowing again otherwise by target in n camera
Mark is together in series, and cluster obtains target trajectory;
Track determining module, the target trajectory for being obtained according to cluster, determines across the scene track of target;
Track display module shows across the scene track of the target of the determination in the electricity for making electronic map
On sub- map.
The module of target detection specifically includes:
Target detection submodule, for by the way of Machine learning classifiers or deep learning, to the road n video counts
It is detected according to the target of middle appearance;
Target acquistion submodule, for being caught by extracting region candidate and classification to all kinds of targets in each scene
It obtains.
The target tracking module specifically includes:
Tracker constructs module, is each for utilizing twin network and region candidate network struction monotrack device
Target establishes respective monotrack device respectively;
Track generation module obtains the target of the appearance in each camera for extending to multiple target tracking
Track.
Identification module specifically includes the target again:
Target critical point detection module determines the accurate upper and lower, left and right of target for carrying out critical point detection to target
Four boundaries obtain the maximum region of interest ROI of target accounting;
Target's feature-extraction module, for extracting clarification of objective by way of deep learning;
Trajectory clustering module, for the track for belonging to same target in n camera to be together in series, cluster obtains target
Track.
The target's feature-extraction module specifically includes:
First object characteristic extracting module, for extracting clarification of objective from by the way of encoding using unsupervised learning;
Alternatively,
Second target's feature-extraction module, for extracting clarification of objective using convolutional neural networks.
The trajectory clustering module specifically includes:
A bit in mapping, rail are seen for the method according to hierarchical clustering in the track of each target by hierarchical clustering module
The distance between mark and track constitute side, cut mode by the figure of max-flow min-cut, will belong to all tracks of same target
It gets together, obtains target trajectory.
The track determining module specifically includes:
Track association module, the target trajectory for being obtained according to cluster, the time occurred in conjunction with target in each track are suitable
Sequence and spatial relation determine the cross-scenario track of target.
The track display module specifically includes:
Trajectory map module makes electronic map for the region where the n camera that be mutually related according to,
On the electronic map, across the scene track of the target is mapped according to specific location, to show target entire
The running route in region.
The device have association multi-cam monitoring scene in, by target each single camera capture after, benefit
Target is again identified that with the hierarchical clustering of track, same target is connected into cross-scenario rail in the track of single scene
Mark realizes the tracking to target under across scene.
Referring to fig. 4, Fig. 4 is a kind of schematic diagram of Computer Vision Platform provided in an embodiment of the present invention;As shown in figure 4,
A kind of Computer Vision Platform provided in an embodiment of the present invention, comprising: n camera and a kind of above-mentioned target that is based on identify again
Across scene follow-up mechanism.For the specific structure across scene follow-up mechanism identified again based on target, referring to aforementioned reality
Apply the specific descriptions of example.
The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention.
Various modifications to these embodiments will be readily apparent to those skilled in the art, as defined herein
General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, of the invention
It is not intended to be limited to the embodiments shown herein, and is to fit to and the principles and novel features disclosed herein phase one
The widest scope of cause.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (17)
1. a kind of across scene method for tracing identified again based on target characterized by comprising
The video data for acquiring the n camera that be mutually related in real time, obtains the road n video data;
The target occurred in the video data of the road n is detected, the target occurred in each scene is captured;
The target occurred in each scene of the capture is tracked in subsequent frames, obtains the target of the appearance every
Track in a camera;
Known again by target otherwise, the track that same target is belonged in n camera is together in series, cluster obtains target
Track;
According to the target trajectory that cluster obtains, across the scene track of target is determined;
Electronic map is made, across the scene track of the target of the determination is shown on the electronic map.
2. the method according to claim 1, wherein it is described to the target occurred in the video data of the road n into
Row detection, captures the target occurred in each scene, specifically includes:
By the way of Machine learning classifiers or deep learning, the target occurred in the video data of the road n is detected,
All kinds of targets in each scene are captured by extracting region candidate and classification.
3. the method according to claim 1, wherein the target occurred in each scene to the capture
It is tracked in subsequent frames, obtains track of the target of the appearance in each camera, specifically include:
Using twin network and region candidate network struction monotrack device, respective list is then established respectively for each target
Target tracker, and then multiple target tracking is extended to, to obtain track of the target of the appearance in each camera.
4. the method according to claim 1, wherein described known otherwise again by target, by n camera
In belong to the track of same target and be together in series, cluster obtains target trajectory, specifically include:
Critical point detection is carried out to target, four, accurate upper and lower, left and right of target boundary is determined, it is maximum to obtain target accounting
Region of interest ROI;
Clarification of objective is extracted by way of deep learning;
The track for belonging to same target in n camera is together in series, cluster obtains target trajectory.
5. according to the method described in claim 4, it is characterized in that, the spy for extracting target by way of deep learning
Sign, specifically includes:
Clarification of objective is extracted from by the way of encoding using unsupervised learning;
Alternatively,
Clarification of objective is extracted using convolutional neural networks.
6. according to the method described in claim 4, it is characterized in that, the track that same target will be belonged in n camera
It being together in series, cluster obtains target trajectory, it specifically includes:
According to the method for hierarchical clustering, a bit in mapping, the distance between track and track structure are seen into the track of each target
Cheng Bian cuts mode by the figure of max-flow min-cut, and all tracks for belonging to same target are got together, target track is obtained
Mark.
7. the method according to claim 1, wherein the target trajectory obtained according to cluster, determines target
Across scene track, specifically includes:
According to the target trajectory that cluster obtains, the time sequencing occurred in conjunction with target in each track and spatial relation are determined
The cross-scenario track of target.
8. the method according to claim 1, wherein the production electronic map, by the target of the determination across
Scene track is shown on the electronic map, is specifically included:
Region where the described n camera that be mutually related makes electronic map, will be described on the electronic map
Across the scene track of target is mapped according to specific location, to show target in the running route of whole region.
9. a kind of across scene follow-up mechanism identified again based on target characterized by comprising
Video acquisition module obtains the road n video data for acquiring the video data of the n camera that be mutually related in real time;
Module of target detection captures and occurs in each scene for detecting to the target occurred in the video data of the road n
Target;
Target tracking module, the target for occurring in each scene to the capture are tracked in subsequent frames, are obtained
Track of the target of the appearance in each camera;
Target again go here and there the track for belonging to same target in n camera for being known again otherwise by target by identification module
Connection gets up, and cluster obtains target trajectory;
Track determining module, the target trajectory for being obtained according to cluster, determines across the scene track of target;
Track display module, for making electronic map, by across the scene track of the target of the determination show it is described electronically
On figure.
10. device according to claim 9, which is characterized in that the module of target detection specifically includes:
Target detection submodule, for by the way of Machine learning classifiers or deep learning, in the video data of the road n
The target of appearance is detected;
Target acquistion submodule, for being captured by extracting region candidate and classification to all kinds of targets in each scene.
11. device according to claim 9, which is characterized in that the target tracking module specifically includes:
Tracker constructs module, is each target for utilizing twin network and region candidate network struction monotrack device
Respective monotrack device is established respectively;
Track generation module obtains track of the target of the appearance in each camera for extending to multiple target tracking.
12. device according to claim 9, which is characterized in that identification module specifically includes the target again:
Target critical point detection module determines four, the accurate upper and lower, left and right of target for carrying out critical point detection to target
Boundary obtains the maximum region of interest ROI of target accounting;
Target's feature-extraction module, for extracting clarification of objective by way of deep learning;
Trajectory clustering module, for the track for belonging to same target in n camera to be together in series, cluster obtains target track
Mark.
13. device according to claim 12, which is characterized in that the target's feature-extraction module specifically includes:
First object characteristic extracting module, for extracting clarification of objective from by the way of encoding using unsupervised learning;
Alternatively,
Second target's feature-extraction module, for extracting clarification of objective using convolutional neural networks.
14. device according to claim 12, which is characterized in that the trajectory clustering module specifically includes:
Hierarchical clustering module, for the method according to hierarchical clustering, by the track of each target see in mapping a bit, track with
The distance between track constitutes side, cuts mode by the figure of max-flow min-cut, all tracks for belonging to same target are gathered
Together, target trajectory is obtained.
15. device according to claim 9, which is characterized in that the track determining module specifically includes:
Track association module, for according to the obtained target trajectory of cluster, the time sequencing occurred in conjunction with target in each track and
Spatial relation determines the cross-scenario track of target.
16. device according to claim 9, which is characterized in that the track display module specifically includes:
Trajectory map module makes electronic map for the region where the n camera that be mutually related according to, described
On electronic map, across the scene track of the target is mapped according to specific location, to show target in whole region
Running route.
17. a kind of Computer Vision Platform characterized by comprising n camera and such as any one of claim 9-16 institute
A kind of across the scene follow-up mechanism identified again based on target stated.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810937026.7A CN108986158A (en) | 2018-08-16 | 2018-08-16 | A kind of across the scene method for tracing identified again based on target and device and Computer Vision Platform |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810937026.7A CN108986158A (en) | 2018-08-16 | 2018-08-16 | A kind of across the scene method for tracing identified again based on target and device and Computer Vision Platform |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108986158A true CN108986158A (en) | 2018-12-11 |
Family
ID=64553759
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810937026.7A Pending CN108986158A (en) | 2018-08-16 | 2018-08-16 | A kind of across the scene method for tracing identified again based on target and device and Computer Vision Platform |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108986158A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109711316A (en) * | 2018-12-21 | 2019-05-03 | 广东工业大学 | A kind of pedestrian recognition methods, device, equipment and storage medium again |
CN109872342A (en) * | 2019-02-01 | 2019-06-11 | 北京清帆科技有限公司 | A kind of method for tracking target under special scenes |
CN110472551A (en) * | 2019-08-09 | 2019-11-19 | 视云融聚(广州)科技有限公司 | A kind of across mirror method for tracing, electronic equipment and storage medium improving accuracy |
CN111147778A (en) * | 2019-12-23 | 2020-05-12 | 浙江大华技术股份有限公司 | Track recording method and device based on digital video recorder and storage medium |
CN111300431A (en) * | 2020-03-31 | 2020-06-19 | 山东大学 | Cross-scene-oriented robot vision simulation learning method and system |
CN111681149A (en) * | 2020-06-05 | 2020-09-18 | 视云融聚(广州)科技有限公司 | Nursing home guardianship object refined monitoring method, storage device and mobile terminal |
CN111784729A (en) * | 2020-07-01 | 2020-10-16 | 杭州海康威视数字技术股份有限公司 | Object tracking method and device, electronic equipment and storage medium |
CN111784730A (en) * | 2020-07-01 | 2020-10-16 | 杭州海康威视数字技术股份有限公司 | Object tracking method and device, electronic equipment and storage medium |
CN112200841A (en) * | 2020-09-30 | 2021-01-08 | 杭州海宴科技有限公司 | Cross-domain multi-camera tracking method and device based on pedestrian posture |
CN112560621A (en) * | 2020-12-08 | 2021-03-26 | 北京大学 | Identification method, device, terminal and medium based on animal image |
WO2021114702A1 (en) * | 2019-12-10 | 2021-06-17 | 中国银联股份有限公司 | Target tracking method, apparatus and system, and computer-readable storage medium |
CN113112726A (en) * | 2021-05-11 | 2021-07-13 | 创新奇智(广州)科技有限公司 | Intrusion detection method, device, equipment, system and readable storage medium |
CN115618250A (en) * | 2022-12-02 | 2023-01-17 | 华清瑞达(天津)科技有限公司 | Radar target obstacle simulation and identification method |
-
2018
- 2018-08-16 CN CN201810937026.7A patent/CN108986158A/en active Pending
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109711316B (en) * | 2018-12-21 | 2022-10-21 | 广东工业大学 | Pedestrian re-identification method, device, equipment and storage medium |
CN109711316A (en) * | 2018-12-21 | 2019-05-03 | 广东工业大学 | A kind of pedestrian recognition methods, device, equipment and storage medium again |
CN109872342A (en) * | 2019-02-01 | 2019-06-11 | 北京清帆科技有限公司 | A kind of method for tracking target under special scenes |
CN110472551A (en) * | 2019-08-09 | 2019-11-19 | 视云融聚(广州)科技有限公司 | A kind of across mirror method for tracing, electronic equipment and storage medium improving accuracy |
WO2021114702A1 (en) * | 2019-12-10 | 2021-06-17 | 中国银联股份有限公司 | Target tracking method, apparatus and system, and computer-readable storage medium |
CN111147778A (en) * | 2019-12-23 | 2020-05-12 | 浙江大华技术股份有限公司 | Track recording method and device based on digital video recorder and storage medium |
CN111300431A (en) * | 2020-03-31 | 2020-06-19 | 山东大学 | Cross-scene-oriented robot vision simulation learning method and system |
CN111300431B (en) * | 2020-03-31 | 2021-04-23 | 山东大学 | Cross-scene-oriented robot vision simulation learning method and system |
CN111681149A (en) * | 2020-06-05 | 2020-09-18 | 视云融聚(广州)科技有限公司 | Nursing home guardianship object refined monitoring method, storage device and mobile terminal |
CN111784729A (en) * | 2020-07-01 | 2020-10-16 | 杭州海康威视数字技术股份有限公司 | Object tracking method and device, electronic equipment and storage medium |
CN111784730A (en) * | 2020-07-01 | 2020-10-16 | 杭州海康威视数字技术股份有限公司 | Object tracking method and device, electronic equipment and storage medium |
CN111784729B (en) * | 2020-07-01 | 2023-09-05 | 杭州海康威视数字技术股份有限公司 | Object tracking method and device, electronic equipment and storage medium |
CN112200841A (en) * | 2020-09-30 | 2021-01-08 | 杭州海宴科技有限公司 | Cross-domain multi-camera tracking method and device based on pedestrian posture |
CN112560621A (en) * | 2020-12-08 | 2021-03-26 | 北京大学 | Identification method, device, terminal and medium based on animal image |
CN113112726A (en) * | 2021-05-11 | 2021-07-13 | 创新奇智(广州)科技有限公司 | Intrusion detection method, device, equipment, system and readable storage medium |
CN115618250A (en) * | 2022-12-02 | 2023-01-17 | 华清瑞达(天津)科技有限公司 | Radar target obstacle simulation and identification method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108986158A (en) | A kind of across the scene method for tracing identified again based on target and device and Computer Vision Platform | |
CN105745687B (en) | Context aware Moving target detection | |
CN110363140B (en) | Human body action real-time identification method based on infrared image | |
Du et al. | The unmanned aerial vehicle benchmark: Object detection and tracking | |
CN108062349B (en) | Video monitoring method and system based on video structured data and deep learning | |
CN107943837B (en) | Key-framed video abstract generation method for foreground target | |
US9299162B2 (en) | Multi-mode video event indexing | |
CN102722698B (en) | Method and system for detecting and tracking multi-pose face | |
CN104091176B (en) | Portrait comparison application technology in video | |
WO2019083738A9 (en) | Methods and systems for applying complex object detection in a video analytics system | |
CN104303193B (en) | Target classification based on cluster | |
CN109145708B (en) | Pedestrian flow statistical method based on RGB and D information fusion | |
CN105187785B (en) | A kind of across bayonet pedestrian's identifying system and method based on choice of dynamical notable feature | |
CN103295016B (en) | Behavior recognition method based on depth and RGB information and multi-scale and multidirectional rank and level characteristics | |
CN103164858A (en) | Adhered crowd segmenting and tracking methods based on superpixel and graph model | |
CN103473530A (en) | Adaptive action recognition method based on multi-view and multi-mode characteristics | |
CN105528794A (en) | Moving object detection method based on Gaussian mixture model and superpixel segmentation | |
CN102521565A (en) | Garment identification method and system for low-resolution video | |
CN112200841B (en) | Cross-domain multi-camera tracking method and device based on pedestrian posture | |
CN105160649A (en) | Multi-target tracking method and system based on kernel function unsupervised clustering | |
CN103279791A (en) | Pedestrian counting method based on multiple features | |
CN110781806A (en) | Pedestrian detection tracking method based on YOLO | |
CN103593679A (en) | Visual human-hand tracking method based on online machine learning | |
CN107657232A (en) | A kind of pedestrian's intelligent identification Method and its system | |
CN104301585A (en) | Method for detecting specific kind objective in movement scene in real time |
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: 20181211 |
|
RJ01 | Rejection of invention patent application after publication |