Gymnastic Posture Detection Based on Deep Learning
Wu Wen, Yong Jian Yang, Jingyi Du, Lixiang Liu, Jiahao Wang · 2019
Attitude detection can help the gymnast's posture movements correct. The general method is to extract the contour of the moving target in a video frame, but this method has poor real-time performance and low accuracy. To this end, this paper proposes a deep learning real-time attitude detection method to detect the posture of gymnasts. Input an image, extract features through the convolutional network, correctly link the detected key points of the human body in the picture, and finally merge them into one's overall skeleton to detect the posture of the gymnast. The method has good real-time performance and accuracy high.