Pedestrian re-identification method and system based on key point optimization and multi-hop attention graph convolution

Abstract

The invention proposes a pedestrian re-identification method and system based on key point optimization and multi-hop attention graph convolution, and the method comprises the steps: extracting key points in an image, obtaining a confidence level corresponding to each key point, and obtaining a pedestrian re-identification result based on the low confidence level of the corresponding key point when a pedestrian in the image is shielded; approximately estimating feature information of the shielded key points by using corresponding features of other key points for the key points with low confidence levels; for the processed feature information of the key points, using an added feature learning network with a multi-hop attention mechanism for learning, considering effective information of each layer in the network, combining with an attention matrix with the feature information to acquire output of feature learning; and for the output features, enabling the feature matching network to compare the node relationship in the two images through a graph matching algorithm and the human body topological features, and output a predicted classification result. According to the invention, the precision of re-identification of shielded pedestrians is improved.

Description

technical field [0001] The invention belongs to the technical field of image recognition, and in particular relates to a pedestrian re-identification method and system based on key point optimization and multi-hop attention map convolution. Background technique [0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art. [0003] The goal of person re-identification is to find a specified person among multiple uncorrelated cameras. Most of the existing methods use deep learning methods to obtain human appearance features by training neural networks. These methods may achieve good results on several specific data sets, but in actual situations due to camera The captured images may not construct an ideal data set. Specifically, due to the influence of external environmental factors, pedestrians may be under the shadow of various obstacles, and it is difficult to obtain a complete p...

AI Technical Summary

Problems solved by technology

[0005] However, the inventor found in the research that although there are many solutions to the occluded pedestrian re-identification problem, most of them adopt the method of directly discarding the occlusion points. For example, some body parts below the input image are occluded, which leads to The learning effect of the model on this part of the occlusion is very poor, so when some parts are occluded and the learning significance of the model is not outstanding for this part, these parts will be deliberately discarded, and only those parts that are not occluded will be learned

Clearly, this strategy is not reliable

Images