2021/04/22 by Zubair Martin, Martin, Zubair, Amir Patel +3 · 1 citation
Computer Science · #Anomaly Detection Techniques and Applications #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Hand Gesture Recognition Systems #Human Pose and Action Recognition
paper · pdf · doi:10.48550/arxiv.2104.10916
openalex publication_date 2021/04/22 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Video analysis in tackle-collision based sports is highly subjective and\nexposed to bias, which is inherent in human observation, especially under time\nconstraints. This limitation of match analysis in tackle-collision based sports\ncan be seen as an opportunity for computer vision applications. Objectively\ntracking, detecting and recognising an athlete's movements and actions during\nmatch play from a distance using video, along with our improved understanding\nof injury aetiology and skill execution will enhance our understanding how\ninjury occurs, assist match day injury management, reduce referee subjectivity.\nIn this paper, we present a system of objectively evaluating in-game tackle\nrisk in rugby union matches. First, a ball detection model is trained using the\nYou Only Look Once (YOLO) framework, these detections are then tracked by a\nKalman Filter (KF). Following this, a separate YOLO model is used to detect\npersons/players within a tackle segment and then the ball-carrier and tackler\nare identified. Subsequently, we utilize OpenPose to determine the pose of\nball-carrier and tackle, the relative pose of these is then used to evaluate\nthe risk of the tackle. We tested the system on a diverse collection of rugby\ntackles and achieved an evaluation accuracy of 62.50%. These results will\nenable referees in tackle-contact based sports to make more subjective\ndecisions, ultimately making these sports safer.\n