Pedestrian re-identification system and method

Abstract

The invention provides a pedestrian re-identification system and method, and belongs to the technical field of pedestrian re-identification. The system comprises the following steps of carries out image reconstruction on different original images through a deep learning network based on sparse coding to obtain corresponding reconstruction matrixes; extracting a feature vector in each reconstruction matrix in combination with an attention mechanism; calculating a classification loss result and a verification loss result of the feature vector; and judging whether the feature extraction module converges or not according to the classification loss result and the verification loss result, if yes, calculating a difference degree between the feature vectors of different reconstruction matrixes, if the difference degree is greater than a set threshold value, determining that the feature vectors do not belong to the same pedestrian, and if the difference degree is less than the set threshold value, determining that the feature vectors belong to the same pedestrian. According to the invention, the reconstructed sub-network is used to reconstruct the image so as to improve the image definition, and the multi-task loss function is used to shorten the distance between the same individuals, thereby improving the feature representation capability and discrimination capability of the network,and improving the pedestrian re-identification accuracy.

Description

technical field [0001] The invention relates to the technical field of pedestrian re-identification, in particular to a pedestrian re-identification system and method based on image reconstruction, attention mechanism and multi-task loss function. Background technique [0002] With the development of deep learning, neural network technology is applied in more and more scenarios, and person re-identification, as a popular research direction in the field of computer vision, is also receiving more and more attention. At present, the research on person re-identification is mainly based on representation learning and metric learning. Because the neural network can automatically extract the representational features from the original image data according to the task requirements, many scholars regard the pedestrian re-identification problem as a classification problem or a verification problem, and use a single branch network or a twin network to extract pedestrian features. Alth...

AI Technical Summary

Problems solved by technology

[0005] However, the above implementation scheme still has the following defects: because the network extracts global features and does not focus on features with a large contribution, the image features extracted by the neural network have the same contribution, and no corresponding enhancement and Weakening, low accuracy; at the same time, the loss function of the classification task and the verification task uses cross-entropy loss, which cannot evaluate the distance between features, resulting in low re-identification accuracy; the image captured by the camera has low pixels and the outline of pedestrians is not clear , directly extracting the features of the original image can easily lead to classification errors

Images