Pedestrian attribute recognition method and system based on attribute feature learning decoupling

Abstract

The invention belongs to the fields of pattern recognition, computer vision, visual scene analysis and multi-label classification, and specifically relates to a pedestrian attribute recognition method and system based on attribute feature learning and decoupling, aiming to solve the problem of using the same feature to classify different attributes in existing technologies , so that the effectiveness of pedestrian attribute recognition is low and the robustness is not strong. The present invention includes: through the feature extraction model constructed based on the deep neural network, and extracting the convolution image features of the pre-processed image to be recognized; preset learnable parameters and obtain the attribute index feature of each category attribute; through semantic space mutual attention The module extracts attribute features and index attention maps; the output of the previous semantic space mutual attention module is used as the input of the current module for iteration; the final attribute feature classification of the image to be recognized is obtained by iteration through the attribute classifier. The invention can be applied to attribute recognition of pedestrian pictures in various scenes, and can significantly improve the performance of pedestrian picture attribute recognition.

Description

technical field [0001] The invention belongs to the field of pattern recognition, computer vision, visual scene analysis and multi label classification, in particular to a pedestrian attribute recognition method and system based on attribute feature learning decoupling. Background technology [0002] In recent years, computer vision, artificial intelligence, machine perception and other fields have developed rapidly. With the wide deployment of security cameras, how to carry out efficient pedestrian attribute recognition in the monitoring scene has attracted extensive attention. Pedestrian attribute recognition in the monitoring scene is to use computer algorithms to process and analyze the pedestrian pictures in the video, and automatically obtain the attribute categories of a pedestrian, such as age, gender, backpack, clothing and so on. So as to provide support and assistance for downstream pedestrian image retrieval and pedestrian re recognition technology. [0003] Tradition...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the above-mentioned problems in the prior art, that is, the existing technology uses the same feature to classify different attributes, so that the effectiveness and robustness of pedestrian attribute recognition are low, the present invention provides an attribute-based feature learning A decoupled pedestrian attribute recognition method, the method includes:

Images