Multi-style pedestrian re-identification method, system and terminal based on adversarial learning
A pedestrian re-identification and multi-style technology, applied in the field of computer vision, can solve the problems of high-level features lacking the underlying details of original data, limiting practicability, and not being able to solve the modal gap well
Inactive Publication Date: 2020-11-24
SHANGHAI JIAO TONG UNIV
View PDF2 Cites 2 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
They usually require pre-trained feature extractors for good performance, which limits their usefulness
In addition, adversarial learning on the feature plane cannot well address the gap between modalities, since the high-level features always lack the low-level details of the original data
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0067] The following is a detailed description of the embodiments of the present invention: this embodiment is implemented on the premise of the technical solution of the present invention, and provides detailed implementation methods and specific operation processes. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present invention, and these all belong to the protection scope of the present invention.
[0068] Embodiments of the present invention provide a multi-style pedestrian re-identification method based on adversarial learning, such as figure 1 As shown, the method includes:
[0069] Construct the image generator G, input the original image I={a,b} obtained from the source of the virtual style image and the real image into the modality-invariant image generator G data space, and obtain the mapped output image Among them, a and b respectively represent pictures belonging to diff...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention provides a multi-style pedestrian re-identification method, system and terminal based on adversarial learning. The system comprises a data transformation and classification recognition function module, an image generator serves as a data space transformation function module, data from different sources is transferred to a modal invariant space, and the problem of inconsistency causedby modal differences in multi-style pedestrian pictures is eliminated. Meanwhile, adversarial learning between a classification recognizer and a modal recognizer is used for guiding representation learning; the modal recognizer distinguishes a real image from a virtual image so as to guide spatial transformation of data to further bridge modal differences, and the classification recognizer is used for final recognition and classification so as to learn invariant features with higher discrimination. The extracted features have higher discrimination and higher robustness, modal differences canbe eliminated at the same time, the performance of an existing feature learning network can be improved, and the matching problem between pedestrian pictures adapting to multiple styles can be bettersolved.
Description
technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a multi-style pedestrian re-identification method, system, and terminal based on adversarial learning. Background technique [0002] Pedestrian re-identification is a key task in intelligent video surveillance and has been a research hotspot in the field of computer vision in recent years. It is suitable for technical fields such as security and public places. Pedestrian re-identification can be defined as: in a non-overlapping video surveillance network, for a given person in a camera, determine whether it appears in other cameras or not. It is an automatic target recognition technology that can quickly locate human targets of interest in the surveillance network, and is an important step in applications such as intelligent video surveillance and human behavior analysis. [0003] With the rapid development of multimedia field and computer vision technology, ...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04
CPCG06V40/25G06V40/10G06N3/045G06F18/2415
Inventor 杨华陈琳
Owner SHANGHAI JIAO TONG UNIV