Real-time basketball shoot and goal event detection based on Deep YOLO network

Extracting statistics of sport players are very beneficial for coaches and players while exercising. Form the other side, deep neural networks has been successfully applied to solve various problems from big data analytics to neural language understanding and computer vision tasks. In this paper we address the problem of shoot and goal detection in basketball games with deep learning and make three main contributions. First we make an image dataset for hoop detection, second we fine-tuned Yolo-v۳ deep network to detect hoop accurately. Third we present a real time method to detect shoot and goal in basketball game. All related codes and dataset and demo videos are publicly available in www.github.com/nvdai/hoop_goal_detection.

کلیدواژه ها:

Shoot detection ، Goal detection ، hoop detection ، YOLO ، Deep Neural Network