Cross-modal person re-identification method based on self-imitation mutual distillation

Abstract

A cross-modal pedestrian re-identification method based on self-imitation and mutual distillation, which involves the field of image processing. Aiming at the deficiency that the existing one-stage feature registration method ignores the difference in feature distribution between modalities and modalities, a two-stage feature registration method is proposed to improve the performance of cross-modal pedestrian re-identification. Its two-stage feature registration includes: 1) Intra-modal feature registration: obtain the prototype features of each pedestrian category in a self-simulation learning method, and achieve this by improving the similarity between all samples of this category and the prototype features Feature registration within a modality; 2) Feature registration between modalities: The mutual distillation learning method is used to reduce the difference in sample distribution of different modalities of the same category. Improve the discriminativeness of features. All samples from two different modalities with the same ID learn each other's feature distribution, thereby reducing the feature difference between modalities. It can be used for intelligent video surveillance, pedestrian tracking and behavior analysis, intelligent security, etc.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a cross-modal pedestrian re-identification method based on self-imitation and mutual distillation, which can be used for intelligent video surveillance, pedestrian tracking and behavior analysis, intelligent security and the like. Background technique [0002] Cross-modal person re-identification has received extensive attention in recent years because of its application prospects and practical application value, and many excellent algorithms have emerged. These algorithms can be roughly divided into three categories: feature registration-based cross-modality person re-identification algorithms, image-generation-based cross-modality person re-identification algorithms, and metric learning-based cross-modality person re-identification algorithms. Compared with the other two types of algorithms, the cross-modal person re-identification algorithm based on feature registration has rec...

AI Technical Summary

Problems solved by technology

In addition to increasing the complexity of the model, this type of method ignores the impact of redundant information within the modal on the accuracy of cross-modal pedestrian retrieval, and only directly performs one-stage feature registration, which cannot effectively alleviate the impact of differences between modalities. Barriers to Performance Improvement

Images