Pedestrian attribute identification method and system in monitoring scene

Abstract

The invention relates to a pedestrian attribute identification method and system in a monitoring scene. The attribute identification method comprises the following steps: obtaining a to-be-detected pedestrian image in the monitoring scene; preprocessing the to-be-detected pedestrian image to obtain a processed image; obtaining convolution image features of the to-be-detected pedestrian image through a deep neural network; determining a weight parameter of each attribute classifier according to a full connection layer and the convolution image features; determining network attribute values of the to-be-detected pedestrian image under different attribute classifiers based on the convolution image features and the weight parameters; determining a predicted value of a corresponding attribute based on each network attribute value; and determining the attribute type of the to-be-detected pedestrian image according to each prediction value. Convolutional image features of the to-be-detected pedestrian image are extracted through the deep neural network, and the weight parameter of each attribute classifier is determined; and network attribute values under different attribute classifiers are obtained, and prediction values of corresponding attributes are also obtained so as to accurately determine the attribute type of the to-be-detected pedestrian image.

Description

technical field [0001] The invention relates to the technical field of visual scene processing and analysis, in particular to a method and system for identifying attributes of pedestrians in a monitoring scene. Background technique [0002] In recent years, fields such as computer vision, artificial intelligence, and machine perception have developed rapidly. With the widespread deployment of cameras, how to perform efficient pedestrian attribute recognition in surveillance scenes has received extensive attention. [0003] Pedestrian attribute recognition in surveillance scenes is to use computer algorithms to process and analyze pedestrian pictures in video, and automatically obtain the attribute categories contained in a certain pedestrian, such as age, gender, backpack, clothing, etc. So as to provide support and assistance for the downstream pedestrian image retrieval and pedestrian re-identification technology. [0004] However, the traditional method obtains the feat...

AI Technical Summary

Problems solved by technology

[0005] Although there have been a lot of previous works that have significantly improved the performance of pedestrian attribute recognition by learning more discriminative visual feature representations and better modeling the relationship between attributes, each method inevitably increases the number of models. The amount of parameters and the complexity of calculation increase the difficulty of pedestrian attribute identification

Images