An improved deep neural network method for an athlete's human motion posture recognition
Zhe Dong, Xiongying Wang · International Journal of Information and Communication Technology · 2022
Aiming at improving the accuracy of motion gesture recognition and reducing the time-consuming recognition, this paper proposes an athlete's body motion gesture recognition method based on improved deep neural network. According to the movement characteristics and basic structural characteristics of the human body, the movement posture recognition standard is designed and set in the model. Collect athletes' human motion images, and perform motion image extraction, residual compensation, filtering, normalisation, and morphological processing. On this basis, the features of human motion images are extracted and fused. The recognition classifier is constructed based on the improved deep neural network, the fused feature vector is input into the classifier, and the recognition result is output. The results of the comparative experiment show that the proposed method has a high recognition rate and short overhead time only at 1.2 s.