Aiming at the problems of high complexity of network model,large number of parameters and slow speed of training and testing in gait recognition,a gait recognition method based on knowl-edge distillation is proposed.The ConvNext-KD model was trained by knowledge distillation method,and the recognition accuracy of ConvNext-KD model was improved without increasing the new training data set,model complexity and model parameter number.The method is simulated in CASIA-B and CASIA-C databases of Chinese Academy of Sciences.The results show that the ConvNext-KD model can significantly shorten the duration of the training test and obtain higher recognition accuracy while keeping the number of parameters and complexity low.
维普
万方数据