Effect of Including Attention Mechanisms on (yolov7) Ability to Recognize Gymnastics Motions

The utilization of deep learning techniques in scientific journals has significantly enhanced the attained outcomes, particularly in the domains of medicine, agriculture, sports, and others. This article introduces a technique for detecting the movements of gymnastics players by incorporating attention mechanisms into the deep learning model (yolov7). The attention is specifically added to the backbone, which is responsible for extracting features and plays a crucial role in subsequent stages of object detection. The scSE attention mechanism was utilized, and the acquired results were compared with various attention mechanisms and with a deep learning model (yoloV7)without any modifications. By implementing attention mechanisms, it was evident through practical experimentation that the accuracy of the model improved compared to using the model without any additional components.