Cross-view gait recognition method and system based on decoupling and identification combined learning

The invention relates to a cross-view gait recognition method and system based on decoupling and identification combined learning, and belongs to the technical field of deep learning and pattern recognition. According to the method, the pedestrian gait features are decoupled through an information bottleneck theory. Compared with a conventional decoupling method of artificially designing a loss function, the method is based on the guidance of an information bottleneck theory, and has reasonable interpretability; besides, a density ratio technique and an adversarial training mode are adopted, so that the model can adaptively learn how to effectively decouple identity features and style features, and the weight of a loss function does not need to be manually adjusted or a complex decoupling rule does not need to be designed. Generally speaking, the decoupling method based on the information bottleneck theory provides a more theoretically complete and adaptive mode to deal with the decoupling problem of the pede