Basketball Action Recognition Method of Deep Neural Network Based on Dynamic Residual Attention Mechanism

Aiming at the problem that features extracted from original C3D (Convolutional 3D) convolutional neural network(C3D) were insufficient, and it was difficult to focus on keyframes, which led low accuracy of basketball players’ action recognition; hence, a recognition method deep network based dynamic residual attention mechanism proposed. Firstly, traditional is improved convolution extract sufficient feature information. Secondly, information selected by obtain key video frames. Finally, algorithm compared with in order demonstrate advance applicability algorithm. Experimental results show this can effectively recognize posture, average posture more than 97%.