Cross-modal pedestrian re-identification method and system based on distribution space alignment

A pedestrian re-identification and spatial alignment technology, applied in the field of computer vision, can solve problems such as difficult tasks, high cost, and large differences in visible light images, and achieve the effect of improving anti-interference ability and recognition ability, and reducing distribution distance.

Pending Publication Date: 2022-04-29
SUN YAT SEN UNIV
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0002] Single-modal pedestrian re-identification based on visible light is only suitable for scenes with sufficient light. When the light is insufficient, infrared cameras need to be used to take infrared pedestrian pictures. Therefore, it is necessary to solve the problem of visible light-infrared cross-modal pedestrian re-identification, and it is necessary to face visible light images. Problems with large differences from infrared images
And the supervised model requires a large number of cross-modal data sets manually labeled, and the cost of labeling cross-modal pedestrian identities is very high compared with single-modal, and the task is more difficult

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Cross-modal pedestrian re-identification method and system based on distribution space alignment
  • Cross-modal pedestrian re-identification method and system based on distribution space alignment
  • Cross-modal pedestrian re-identification method and system based on distribution space alignment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0053] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. For the step numbers in the following embodiments, it is only set for the convenience of illustration and description, and the order between the steps is not limited in any way. The execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art sexual adjustment.

[0054]The weakly supervised visible light-infrared cross-modal pedestrian re-identification of the present invention is a weakly supervised and cross-domain retrieval task, which is decomposed into two-stage tasks: the self-supervised stage is trained and fitted by "clustering-fine-tuning" The feature extractor of two modal pedestrian data sets, and use a small amount of data with real annotations to modify the model, learn the view-invariant feature representation and put pseudo-labels on the training data; The outp...

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

No PUM Login to view more

Abstract

The invention discloses a cross-modal pedestrian re-identification method and system based on distribution space alignment. The method comprises the following steps: constructing a feature extractor and training the feature extractor based on a first training sample; a modal image is acquired, and pedestrian modal features are extracted based on the trained feature extractor; clustering the to-be-detected images according to the modal features of the pedestrians; constructing a modal sharing model, and training the modal sharing model based on the data set with the pseudo label; training a first modal sharing model based on the second training sample; and identifying an input image according to the trained modal sharing model. The system comprises a first construction module, a feature extraction module, a clustering module, a second construction module and an identification module. By using the method of the invention, a cross-modal pedestrian re-identification task can be carried out in two modes of visible light and infrared light. The cross-modal pedestrian re-identification method and system based on distribution space alignment can be widely applied to the field of computer vision.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a cross-modal pedestrian re-identification method and system based on distribution space alignment. Background technique [0002] Single-modal pedestrian re-identification based on visible light is only suitable for scenes with sufficient light. When the light is insufficient, infrared cameras need to be used to take infrared pedestrian pictures. Therefore, it is necessary to solve the problem of visible light-infrared cross-modal pedestrian re-identification, and it is necessary to face visible light images. Problems with large differences from infrared images. Moreover, the supervised model requires a large number of manually labeled cross-modal data sets, and the cost of cross-modal pedestrian identity labeling is much higher than that of single-modality, and the task is more difficult. Contents of the invention [0003] In order to solve the above technical problems, the obj...

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
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06V20/52G06V40/10G06V10/143G06V10/74G06V10/762G06V10/764G06V10/774G06V10/82G06N3/04G06N3/08G06K9/62
CPCG06N3/088G06N3/045G06F18/2155G06F18/23G06F18/22G06F18/24
Inventor 赖剑煌刘伟鹏
Owner SUN YAT SEN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products