Pedestrian re-recognition system and method based on spatial sequence feature learning

Abstract

The invention discloses a pedestrian re-recognition system and method based on spatial sequence feature learning. The pedestrian re-recognition system comprises a Res2Net network, a global feature module and a spatial sequence feature learning module. The global feature module comprises a flat pooling module and a full connection layer module; the Res2Net network is respectively connected with the global feature module and the spatial sequence feature learning module; the spatial sequence feature learning module comprises a convolutional layer, a random mask module, a maximum pooling module and a self-attention module. A spatial sequence feature learning module based on a self-attention mechanism is provided, spatial sequence features in the horizontal direction and the vertical direction are constructed, spatial semantic relation of the spatial sequence features is learned, and effective local features are extracted; a random batch feature erasing training strategy is put forward, and a local area of a feature map is shielded through a random mask block, so that the model is forced to learn suppressed low-frequency local features.

Description

technical field [0001] The invention relates to the field of pedestrian re-identification, in particular to a pedestrian re-identification system and method based on spatial sequence feature learning. Background technique [0002] Pedestrian re-identification is an important research direction in the field of computer vision. With the increasing demand for public security and the popularity of public cameras, the role of pedestrian re-identification technology in the field of intelligent security is becoming more and more important. Traditional pedestrian re-identification research is mainly based on manually constructing pedestrian features. With the development of deep learning technology, the performance of pedestrian re-identification model has been significantly improved, but the resolution of pedestrians is low, the image is occluded, and the size of the data set is small. Factors restrict the improvement of model performance. At this stage, many studies ignore the sp...

AI Technical Summary

Problems solved by technology

[0006] (2) Although the introduction of posture and other information through the method of target detection can help improve the model effect, the process is relatively cumbersome and increases the complexity of the algorithm

Moreover, it is a difficult task to perform high-precision attitude detection on pedestrian images. If wrong attitude information is introduced, it will interfere with the model.

Images