CN112907627A - System, method and device for realizing accurate tracking of small sample target, processor and computer readable storage medium thereof - Google Patents
System, method and device for realizing accurate tracking of small sample target, processor and computer readable storage medium thereof Download PDFInfo
- Publication number
- CN112907627A CN112907627A CN202110176675.1A CN202110176675A CN112907627A CN 112907627 A CN112907627 A CN 112907627A CN 202110176675 A CN202110176675 A CN 202110176675A CN 112907627 A CN112907627 A CN 112907627A
- Authority
- CN
- China
- Prior art keywords
- portrait
- module
- image
- small sample
- accurate 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000004458 analytical method Methods 0.000 claims abstract description 35
- 238000000605 extraction Methods 0.000 claims abstract description 5
- 238000011176 pooling Methods 0.000 claims description 16
- 238000005516 engineering process Methods 0.000 claims description 9
- 239000013604 expression vector Substances 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 7
- 238000001514 detection method Methods 0.000 claims description 5
- 239000000284 extract Substances 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 238000005259 measurement Methods 0.000 claims description 4
- 230000004044 response Effects 0.000 claims description 4
- 238000012163 sequencing technique Methods 0.000 claims description 4
- 239000013598 vector Substances 0.000 claims description 4
- 238000003709 image segmentation Methods 0.000 abstract description 4
- 230000000007 visual effect Effects 0.000 description 10
- 230000008859 change Effects 0.000 description 6
- 238000005286 illumination Methods 0.000 description 4
- 230000036544 posture Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/30196—Human being; Person
Abstract
The invention relates to a system for realizing accurate tracking of a small sample target, which comprises a portrait acquisition module, a portrait view acquisition module and a portrait tracking module, wherein the portrait acquisition module is used for acquiring portrait view information from a video; the portrait access convergence module is used for accessing and forwarding the portrait view information and converging the collected portrait view information through the collection interface; the portrait view library is used for storing portrait pictures, portrait video clips and portrait passing records; the portrait analyzing module is used for providing portrait feature extraction and a portrait 1: 1 comparison, 1: n retrieval, N: n clustering, N: comparing and retrieving service of M collision; the portrait application service support module is used for sending a task instruction to the portrait analysis module through the analysis interface. The invention also relates to a method for realizing accurate tracking of the small sample target. By adopting the system, the method, the device, the processor and the computer readable storage medium for realizing the accurate tracking of the small sample target, the external interference caused by complex background and shielding is removed by means of image segmentation and the like, the false alarm rate of the single-dimensional biological feature is reduced, and the method has wider application range.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to the field of visual target tracking, and specifically relates to a system, a method, a device, a processor and a computer readable storage medium for realizing accurate tracking of a small sample target.
Background
Visual Object Tracking (Visual Object Tracking) is an important problem in the field of computer vision, and can realize human body Tracking, face Tracking, vehicle Tracking in traffic monitoring systems, gesture Tracking in intelligent interactive systems, automatic target Tracking of unmanned aerial vehicles and the like. Although widely researched in recent years, the target tracking problem is slightly lower than basic visual tasks such as target detection, semantic segmentation and the like due to the high difficulty and the rare high-quality data of the target tracking problem. The development of deep learning and the enhancement of computer computing power bring about a sudden leap and leap forward of the performance of a visual algorithm, and a method based on a deep neural network in the field of target tracking is just about in the last few years, which is probably more than ideal.
The key to realize target tracking is to completely segment the target, reasonably extract features and accurately identify the target, and simultaneously consider the time for realizing the algorithm so as to ensure the real-time property. Due to the characteristics of rigid and flexible objects, the pedestrian has an appearance that is easily affected by various complex factors such as wearing, posture and visual angle changes, illumination, shielding and environment, so that the human body tracking faces a huge technical challenge. The method is limited by the limitation of cross-resource videos, image shooting environments and pedestrian appearances, the accuracy and reliability of visual target tracking in application are low, and the use significance is lost in most practical application scenes.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a system, a method, a device, a processor and a computer readable storage medium thereof for realizing accurate tracking of a small sample target, which meet the requirements of refinement, structuralization, efficient deployment and control and real-time pursuit.
In order to achieve the above object, the system, method, apparatus, processor and computer readable storage medium thereof for implementing accurate tracking of small sample targets of the present invention are as follows:
the system for realizing the accurate tracking of the small sample target is mainly characterized by comprising the following components:
the image acquisition module is used for acquiring image view information from the video;
the portrait access convergence module is connected with the portrait acquisition module and used for accessing and forwarding the portrait view information and converging the acquired portrait view information through the acquisition interface;
the portrait view library is connected with the portrait access convergence module and is used for storing portrait pictures, portrait video clips and passerby records;
the portrait analyzing module is connected with the portrait view library and used for providing portrait feature extraction and portrait 1: 1 comparison, 1: n retrieval, N: n clustering, N: comparing and retrieving service of M collision;
and the portrait application service support module is connected with the portrait view library and the portrait analysis module, and is used for sending a task instruction to the portrait analysis module through an analysis interface and returning a result by the portrait analysis module and the view library.
Preferably, the system further comprises a top and bottom portrait view library, connected with the portrait view library, and networked through a cascade interface, for transmitting top or bottom portrait recognition information.
Preferably, the portrait view library receives and stores the portrait view information sent by the portrait access convergence module and the analysis module; and a portrait picture is provided to the analysis module through the data service interface, and is used for extracting a portrait characteristic value and providing an analysis result to the portrait application support module.
Preferably, the portrait view information includes a portrait video clip, a portrait picture and a past record.
The method for realizing the accurate tracking of the small sample target based on the system is mainly characterized by comprising the following steps:
(1) the portrait acquisition module acquires portrait view information from the video through acquisition equipment and forwards the portrait view information to the portrait access convergence module;
(2) the portrait view library converges the portrait view information transmitted by the portrait access convergence module or the portrait up-and-down view library and performs classified storage;
(3) the portrait application service support module inputs a portrait and issues an instruction of searching a specific portrait or a place sequence to the portrait analysis module;
(4) the human image analyzing module extracts human image features from the human image view library through a pedestrian re-identification technology and judges that a specific pedestrian is detected in an image or video sequence in the human image view library;
(5) the portrait analyzing module returns the location and time information of the snapshot camera to the portrait application service supporting module and sends an instruction, and the portrait view library sends the matched image and video to the portrait application service supporting module.
Preferably, the step (4) specifically includes the following steps:
(4.1) carrying out pedestrian detection to obtain a pedestrian image;
(4.2) cutting the pedestrian image;
(4.3) taking the cut pedestrian images in different view areas as the input of a network, and decomposing the images into different color channel subgraphs for processing respectively;
(4.4) carrying out convolution filtering operation on each sub-image in the convolution layer to obtain the response of different local image blocks as local characteristics;
(4.5) combining all local features to form a feature map as the output of the convolutional layer;
(4.6) performing a maximum pooling operation or an average pooling operation on the generated feature map in the pooling layer;
(4.7) projecting the feature map obtained by the pooling layer to a one-dimensional feature space at the full-connection layer to form a feature vector of the pedestrian image;
(4.8) judging whether the input image pair inputs the same pedestrian or not through a binary function;
(4.9) extracting features aiming at the input image to obtain feature expression vectors for distinguishing different pedestrians;
and (4.10) carrying out similarity measurement according to the feature expression vectors, sequencing the images according to the similarity, and taking the image with the highest similarity as an identification result.
This a device for realizing accurate tracking of small sample target, its key feature is, the device include:
a processor configured to execute computer-executable instructions;
and the memory stores one or more computer-executable instructions, and when the computer-executable instructions are executed by the processor, the steps of the method for realizing the accurate tracking of the small sample target are realized.
The processor for realizing the accurate tracking of the small sample target is mainly characterized in that the processor is configured to execute computer executable instructions, and when the computer executable instructions are executed by the processor, the steps of the method for realizing the accurate tracking of the small sample target are realized.
The computer-readable storage medium is mainly characterized by storing a computer program thereon, wherein the computer program can be executed by a processor to realize the steps of the method for realizing the accurate tracking of the small sample target.
The system, the method, the device, the processor and the computer readable storage medium for realizing the accurate tracking of the small sample target adopt the pedestrian re-identification technology, and adopt the human body to search the image by the image to follow the evolution thinking from a simple background to a complex background, from no occlusion to occlusion, and from small change to large change. External interference caused by complex background and shielding is removed through means such as image segmentation, apparent difference caused by changes of visual angle, resolution, posture, illumination and the like is removed through key point positioning and system robustness improvement, the problem that targets cannot be identified or are identified wrongly under the condition of mass target identification, particularly cross-resource is solved, the false alarm rate of single-dimensional biological characteristics is reduced, and the method has a wider application range.
Drawings
Fig. 1 is a schematic connection diagram of modules of a system for implementing accurate tracking of a small sample target according to the present invention.
Detailed Description
In order to more clearly describe the technical contents of the present invention, the following further description is given in conjunction with specific embodiments.
The system for realizing accurate tracking of the small sample target comprises the following components:
the image acquisition module is used for acquiring image view information from the video;
the portrait access convergence module is connected with the portrait acquisition module and used for accessing and forwarding the portrait view information and converging the acquired portrait view information through the acquisition interface;
the portrait view library is connected with the portrait access convergence module and is used for storing portrait pictures, portrait video clips and passerby records;
the portrait analyzing module is connected with the portrait view library and used for providing portrait feature extraction and portrait 1: 1 comparison, 1: n retrieval, N: n clustering, N: comparing and retrieving service of M collision;
and the portrait application service support module is connected with the portrait view library and the portrait analysis module, and is used for sending a task instruction to the portrait analysis module through an analysis interface and returning a result by the portrait analysis module and the view library.
As a preferred embodiment of the present invention, the system further includes a top and bottom portrait view library, connected to the portrait view library, and networked through a cascade interface, for transmitting top and bottom portrait recognition information.
As a preferred embodiment of the present invention, the portrait view library receives and stores the portrait view information sent by the portrait access convergence module and the portrait analysis module; and a portrait picture is provided to the analysis module through the data service interface, and is used for extracting a portrait characteristic value and providing an analysis result to the portrait application support module.
As a preferred embodiment of the present invention, the portrait view information includes a portrait video clip, a portrait picture and a past record.
The method for realizing the accurate tracking of the small sample target based on the system comprises the following steps:
(1) the portrait acquisition module acquires portrait view information from the video through acquisition equipment and forwards the portrait view information to the portrait access convergence module;
(2) the portrait view library converges the portrait view information transmitted by the portrait access convergence module or the portrait up-and-down view library and performs classified storage;
(3) the portrait application service support module inputs a portrait and issues an instruction of searching a specific portrait or a place sequence to the portrait analysis module;
(4) the human image analyzing module extracts human image features from the human image view library through a pedestrian re-identification technology and judges that a specific pedestrian is detected in an image or video sequence in the human image view library;
(4.1) carrying out pedestrian detection to obtain a pedestrian image;
(4.2) cutting the pedestrian image;
(4.3) taking the cut pedestrian images in different view areas as the input of a network, and decomposing the images into different color channel subgraphs for processing respectively;
(4.4) performing convolution filtering operation on each sub-image in the convolution layer to obtain the response of different local image blocks
Is a local feature;
(4.5) combining all local features to form a feature map as the output of the convolutional layer;
(4.6) performing a maximum pooling operation or an average pooling operation on the generated feature map in the pooling layer;
(4.7) projecting the feature map obtained by the pooling layer to a one-dimensional feature space at the full-connection layer to form a feature vector of the pedestrian image;
(4.8) judging whether the input image pair inputs the same pedestrian or not through a binary function;
(4.9) extracting features aiming at the input image to obtain feature expression vectors for distinguishing different pedestrians;
(4.10) carrying out similarity measurement according to the feature expression vectors, sequencing the images according to the similarity, and taking the image with the highest similarity as an identification result;
(5) the portrait analyzing module returns the location and time information of the snapshot camera to the portrait application service supporting module and sends an instruction, and the portrait view library sends the matched image and video to the portrait application service supporting module.
As a preferred embodiment of the present invention, the apparatus for implementing accurate tracking of a small sample target comprises:
a processor configured to execute computer-executable instructions;
and the memory stores one or more computer-executable instructions, and when the computer-executable instructions are executed by the processor, the steps of the method for realizing the accurate tracking of the small sample target are realized.
As a preferred embodiment of the present invention, the processor for implementing accurate tracking of a small sample target is configured to execute computer-executable instructions, and when the computer-executable instructions are executed by the processor, the steps of the method for implementing accurate tracking of a small sample target are implemented.
As a preferred embodiment of the present invention, the computer readable storage medium has stored thereon a computer program which can be executed by a processor to implement the steps of the above-mentioned method for achieving accurate tracking of a small sample target.
In a specific embodiment of the invention, a system which can be used for completing refined and rapid structural description of a sensitive human target is provided, and by technologies such as online modeling, cross-resource target identification, cross-domain target identification and the like of the sensitive target, the problems that the target cannot be identified or is identified wrongly due to factors such as angle, distance, shielding and the like are solved, and efficient deployment and control and real-time pursuit based on a video target are realized.
The accurate tracking system for the small sample target is shown in figure 1 and comprises a portrait acquisition module, a portrait access convergence module, a portrait view library, a portrait analysis module, an upper portrait view library, a lower portrait view library and a portrait application service support module.
The portrait acquisition module is used for acquiring portrait view information from a video, and comprises the following steps: portrait video clips, portrait pictures, and people-past records.
The portrait access convergence module is used for accessing and forwarding portrait view information, and the collected portrait view information is converged and forwarded to the portrait view library through the collection interface.
The portrait view library mainly comprises functional modules such as an interface, application, management and the like, is used for storing portrait pictures, portrait video clips, passerby records and the like, and is divided into a one-person one-file portrait cluster, a case-related portrait library, an archive special topic library, a control special topic library, a passerby library and the like from the aspects of business and logic.
The portrait parsing module provides portrait feature extraction and portrait 1: 1 comparison, 1: n retrieval, N: n clustering, N: m collision equal ratio retrieval service.
And the portrait view library supports receiving and storing the portrait view information sent by the portrait access convergence module and the portrait analysis module.
The view library provides the portrait pictures for the analysis module through the data service interface for extracting the portrait characteristic values and providing the analysis results for the application support module.
The upper and lower level portrait views are networked through a cascading interface.
The portrait application service support module sends a task instruction to the portrait analysis module through the analysis interface, and the result is returned by the portrait analysis module and the view library.
The method for realizing video target tracking by using the small sample target accurate tracking system based on the system comprises the following steps:
(1) connecting a portrait acquisition module with a portrait access convergence module;
(2) connecting the portrait access convergence module with a portrait view library;
(3) the portrait view library is respectively connected with a portrait analysis module, an upper and lower level portrait view library and a portrait application service support module;
(4) connecting the portrait analyzing module with the portrait application service support module;
(5) the portrait acquisition module acquires portrait view information from the video through acquisition equipment and forwards the portrait view information to the portrait access convergence module;
(6) the portrait access convergence module or the portrait up-and-down view library converges the received portrait view information and forwards the converged portrait view information to the portrait view library for classified storage;
(7) the portrait application service support module inputs a portrait and issues an instruction of searching a specific portrait or a place sequence where the specific portrait appears to the portrait analysis module;
(8) the portrait analyzing module extracts portrait characteristics from the portrait view library by adopting a pedestrian re-identification technology and judges whether specific pedestrians exist in images or video sequences in the view library;
(9) the portrait analyzing module returns the location and time information of the snapshot camera to the portrait application service support module, and sends an instruction to enable the portrait view library to send the matched images and videos to the portrait application service support module.
The pedestrian re-identification technology is a method for removing external interference caused by the complex background and shielding by means of image segmentation and the like and removing apparent difference caused by changes of visual angles, resolution ratios, postures, illumination and the like by positioning key points and improving system robustness, and follows the evolution thinking of searching a picture from a simple background to a complex background, from no shielding to shielding and from small change to large change by a human body with the picture, and the like, and specifically comprises the following steps:
(1) carrying out pedestrian detection to obtain a pedestrian image;
(2) cutting a pedestrian image;
(3) taking the cut pedestrian images in different view areas as the input of a network, and decomposing the images into different color channel subgraphs for processing respectively;
(4) for each sub-image, the convolution filtering operation is carried out in the convolution layer to obtain the response of different local image blocks,
as a local feature;
(5) combining all local features to form a feature map as an output of the convolutional layer;
(6) performing maximum/average pooling operation on the generated feature map in a pooling layer, thereby greatly reducing the occurrence of training parameters and overfitting phenomena;
(7) the convolution layer and the pooling layer can appear for many times to obtain abstract and multi-level description of pedestrians;
(8) projecting the characteristic diagram obtained by the pooling layer to a one-dimensional characteristic space at the full-connection layer to form a characteristic vector of the pedestrian image;
(9) the last Softmax layer judges whether the input image pair is input into the same pedestrian or not through a binary function;
(10) extracting stable and robust features from an input image to obtain feature expression vectors capable of describing and distinguishing different pedestrians;
(11) and finally, carrying out similarity measurement according to the feature expression vectors, sequencing the images according to the similarity, and taking the image with the highest similarity as a final recognition result.
The method realizes the identification, retrieval and discovery of cross-video resources and cross-scene sensitive targets in a typical application environment, solves the problems that the targets are blocked and cannot be identified and tracked when the background is complex through a pedestrian re-identification technology, thereby realizing the automatic association of case and event targets and improving the police affair application efficiency of mass videos of public security organs.
For a specific implementation of this embodiment, reference may be made to the relevant description in the above embodiments, which is not described herein again.
It is understood that the same or similar parts in the above embodiments may be mutually referred to, and the same or similar parts in other embodiments may be referred to for the content which is not described in detail in some embodiments.
It should be noted that the terms "first," "second," and the like in the description of the present invention are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. Further, in the description of the present invention, the meaning of "a plurality" means at least two unless otherwise specified.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and alternate implementations are included within the scope of the preferred embodiment of the present invention in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present invention.
It should be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by suitable instruction execution devices. For example, if implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.
It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware related to instructions of a program, and the corresponding program may be stored in a computer readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.
In addition, functional units in the embodiments of the present invention may be integrated into one processing module, or each unit may exist alone physically, or two or more units are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. The integrated module, if implemented in the form of a software functional module and sold or used as a separate product, may also be stored in a computer readable storage medium.
The storage medium mentioned above may be a read-only memory, a magnetic or optical disk, etc.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.
The system, the method, the device, the processor and the computer readable storage medium for realizing the accurate tracking of the small sample target adopt the pedestrian re-identification technology, and adopt the human body to search the image by the image to follow the evolution thinking from a simple background to a complex background, from no occlusion to occlusion, and from small change to large change. External interference caused by complex background and shielding is removed through means such as image segmentation, apparent difference caused by changes of visual angle, resolution, posture, illumination and the like is removed through key point positioning and system robustness improvement, the problem that targets cannot be identified or are identified wrongly under the condition of mass target identification, particularly cross-resource is solved, the false alarm rate of single-dimensional biological characteristics is reduced, and the method has a wider application range.
In this specification, the invention has been described with reference to specific embodiments thereof. It will, however, be evident that various modifications and changes may be made thereto without departing from the broader spirit and scope of the invention. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense.
Claims (9)
1. A system for realizing accurate tracking of a small sample target is characterized by comprising:
the image acquisition module is used for acquiring image view information from the video;
the portrait access convergence module is connected with the portrait acquisition module and used for accessing and forwarding the portrait view information and converging the acquired portrait view information through the acquisition interface;
the portrait view library is connected with the portrait access convergence module and is used for storing portrait pictures, portrait video clips and passerby records;
the portrait analyzing module is connected with the portrait view library and used for providing portrait feature extraction and portrait 1: 1 comparison, 1: n retrieval, N: n clustering, N: comparing and retrieving service of M collision;
and the portrait application service support module is connected with the portrait view library and the portrait analysis module, and is used for sending a task instruction to the portrait analysis module through an analysis interface and returning a result by the portrait analysis module and the view library.
2. The system for realizing accurate tracking of the small sample target as claimed in claim 1, wherein the system further comprises a portrait view library of upper and lower levels, connected with the portrait view library, networked through a cascade interface, and used for transmitting portrait recognition information of the upper or lower level.
3. The system for realizing accurate tracking of the small sample target according to claim 1, wherein the portrait view library receives and stores portrait view information sent by a portrait access convergence module and an analysis module; and a portrait picture is provided to the analysis module through the data service interface, and is used for extracting a portrait characteristic value and providing an analysis result to the portrait application support module.
4. The system for realizing accurate tracking of the small sample target according to claim 1, wherein the portrait view information comprises a portrait video clip, a portrait picture and a past record.
5. A method for realizing accurate tracking of a small sample target based on the system of claim 1, wherein the method comprises the following steps:
(1) the portrait acquisition module acquires portrait view information from the video through acquisition equipment and forwards the portrait view information to the portrait access convergence module;
(2) the portrait view library converges the portrait view information transmitted by the portrait access convergence module or the portrait up-and-down view library and performs classified storage;
(3) the portrait application service support module inputs a portrait and issues an instruction of searching a specific portrait or a place sequence to the portrait analysis module;
(4) the human image analyzing module extracts human image features from the human image view library through a pedestrian re-identification technology and judges that a specific pedestrian is detected in an image or video sequence in the human image view library;
(5) the portrait analyzing module returns the location and time information of the snapshot camera to the portrait application service supporting module and sends an instruction, and the portrait view library sends the matched image and video to the portrait application service supporting module.
6. The method for achieving accurate tracking of a small sample target according to claim 1, wherein the step (4) specifically comprises the following steps:
(4.1) carrying out pedestrian detection to obtain a pedestrian image;
(4.2) cutting the pedestrian image;
(4.3) taking the cut pedestrian images in different view areas as the input of a network, and decomposing the images into different color channel subgraphs for processing respectively;
(4.4) carrying out convolution filtering operation on each sub-image in the convolution layer to obtain the response of different local image blocks as local characteristics;
(4.5) combining all local features to form a feature map as the output of the convolutional layer;
(4.6) performing a maximum pooling operation or an average pooling operation on the generated feature map in the pooling layer;
(4.7) projecting the feature map obtained by the pooling layer to a one-dimensional feature space at the full-connection layer to form a feature vector of the pedestrian image;
(4.8) judging whether the input image pair inputs the same pedestrian or not through a binary function;
(4.9) extracting features aiming at the input image to obtain feature expression vectors for distinguishing different pedestrians;
and (4.10) carrying out similarity measurement according to the feature expression vectors, sequencing the images according to the similarity, and taking the image with the highest similarity as an identification result.
7. An apparatus for achieving accurate tracking of a small sample target, the apparatus comprising:
a processor configured to execute computer-executable instructions;
a memory storing one or more computer-executable instructions that, when executed by the processor, perform the steps of the method of claim 5 or 6 for achieving accurate tracking of small sample targets.
8. A processor for achieving accurate tracking of a small sample target, wherein the processor is configured to execute computer-executable instructions, which when executed by the processor, implement the steps of the method for achieving accurate tracking of a small sample target as claimed in claim 5 or 6.
9. A computer-readable storage medium, having stored thereon a computer program executable by a processor for performing the steps of the method of claim 5 or 6 for accurate tracking of small sample objects.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110176675.1A CN112907627B (en) | 2021-02-07 | 2021-02-07 | System, method, apparatus, processor and computer readable storage medium for realizing accurate tracking of small sample targets |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110176675.1A CN112907627B (en) | 2021-02-07 | 2021-02-07 | System, method, apparatus, processor and computer readable storage medium for realizing accurate tracking of small sample targets |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112907627A true CN112907627A (en) | 2021-06-04 |
| CN112907627B CN112907627B (en) | 2024-02-02 |
Family
ID=76123090
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110176675.1A Active CN112907627B (en) | 2021-02-07 | 2021-02-07 | System, method, apparatus, processor and computer readable storage medium for realizing accurate tracking of small sample targets |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112907627B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020211003A1 (en) * | 2019-04-17 | 2020-10-22 | 深圳市欢太科技有限公司 | Image processing method, computer readable storage medium, and computer device |
| CN112257502A (en) * | 2020-09-16 | 2021-01-22 | 深圳微步信息股份有限公司 | Pedestrian identification and tracking method and device for surveillance video and storage medium |
-
2021
- 2021-02-07 CN CN202110176675.1A patent/CN112907627B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020211003A1 (en) * | 2019-04-17 | 2020-10-22 | 深圳市欢太科技有限公司 | Image processing method, computer readable storage medium, and computer device |
| CN112257502A (en) * | 2020-09-16 | 2021-01-22 | 深圳微步信息股份有限公司 | Pedestrian identification and tracking method and device for surveillance video and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112907627B (en) | 2024-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Henry et al. | Multinational license plate recognition using generalized character sequence detection | |
| US9443320B1 (en) | Multi-object tracking with generic object proposals | |
| Boult et al. | Into the woods: Visual surveillance of noncooperative and camouflaged targets in complex outdoor settings | |
| US9251425B2 (en) | Object retrieval in video data using complementary detectors | |
| Suprem et al. | Odin: Automated drift detection and recovery in video analytics | |
| CN117095349A (en) | Appearance search system, method, and non-transitory computer readable medium | |
| CN104303193A (en) | Clustering-based object classification | |
| US20230267735A1 (en) | Method for structuring pedestrian information, device, apparatus and storage medium | |
| Varghese et al. | An efficient algorithm for detection of vacant spaces in delimited and non-delimited parking lots | |
| CN110659391A (en) | Video detection method and device | |
| CN111652035B (en) | Pedestrian re-identification method and system based on ST-SSCA-Net | |
| EP3574417A1 (en) | Scalable and efficient episodic memory in cognitive processing for automated systems | |
| Castanon et al. | Retrieval in long-surveillance videos using user-described motion and object attributes | |
| Kwon et al. | Toward an online continual learning architecture for intrusion detection of video surveillance | |
| KR101492059B1 (en) | Real Time Object Tracking Method and System using the Mean-shift Algorithm | |
| Nagrath et al. | Understanding new age of intelligent video surveillance and deeper analysis on deep learning techniques for object tracking | |
| Feris et al. | Case study: IBM smart surveillance system | |
| Liu et al. | Automatic detection of parking violation and capture of license plate | |
| CN115953650B (en) | Training method and device for feature fusion model | |
| Buch et al. | Local feature saliency classifier for real-time intrusion monitoring | |
| CN112907627B (en) | System, method, apparatus, processor and computer readable storage medium for realizing accurate tracking of small sample targets | |
| Negri | Estimating the queue length at street intersections by using a movement feature space approach | |
| CN115393755A (en) | Visual target tracking method, device, equipment and storage medium | |
| US11532158B2 (en) | Methods and systems for customized image and video analysis | |
| CN114092958A (en) | Method, system, electronic device and readable storage medium for object re-identification |
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 |