Sports technology is evolving rapidly, and the National Science and Technology Council has been promoting basic research and innovative technologies in sports technology, encouraging collaboration between academia and industry to develop AI-related systems or products. With the support of the National Science and Technology Council, Professor Min-Chun Hu and her team from the Department of Computer Science at National Tsing Hua University have developed the "AI Sports Video Analysis and VR Immersive Training Platform" to assist athletes in information gathering and training to enhance their skills, tactics, and decision-making abilities. The team's research achievements have been published in several leading international conferences and important international journals such as IEEE Transactions on Multimedia and IEEE Transactions on Visualization and Computer Graphics. They also won the Silver Award and Honorable Mention Award from the Taiwan Sports Technology Association in 2021.
The AI sports video analysis technology developed by the team can detect important events such as shooting and dunking in real-time from videos and analyze athletes' movement techniques and tactics. This helps athletes and sports intelligence personnel efficiently label match video content and decipher opponents' strategies. Additionally, it can reconstruct three-dimensional (3D) information of scenes and objects from match videos, enabling replay from any viewpoint to help athletes create an immersive VR training environment. This aids in enhancing cognitive skills and tactics, thereby improving athletic performance.
By analyzing a vast amount of match videos, the team has developed various generative models. For example, in basketball, it can generate the corresponding defensive movement trajectories based on the attacker's movement trajectories in real-time. It can also generate 3D animations of an opponent's various actions based on past match videos of the athlete's designated opponent, allowing athletes to simulate real-match scenarios more accurately during VR training. Furthermore, the developed gesture recognition technology allows athletes to interact more intuitively with the system during simulation training, also enabling complete recording of athletes' reactions and performances in different situations, which is beneficial for evaluating training effectiveness.
To make AI technology accessible in sports fields for the general public, Professor Hu, along with Professor Hung-Kuo Chu from the Department of Computer Science at National Tsing Hua University, led students to extend the developed technology to establish a startup company (NeuinX). The team has also developed precise and real-time object tracking and shooting event analysis technology for mobile device-captured basketball videos. Basketball enthusiasts can upload or stream their recorded match videos to instantly generate highlights and shooting hotspots. Whether during or after matches, AI technology can assist in quickly gathering meaningful sports data and even produce augmented reality-enhanced highlights for further application in advertising and marketing, adding value to match videos and generating greater commercial revenue. Professor Hu’s team integrated the above-mentioned technology to build “Court Athena”, a sports video analysis and data visualization platform, and was awarded the Innovation Awards at CES 2023!
Beyond basketball, Professor Hu’s research team is also collaborating with the cross-institutional research team from the Sports Technology Center in National Tsing Hua University to further assist in intelligence gathering and training for sports such as table tennis, baseball, boxing, and archery. The realm of sports encompasses a broad spectrum of unique and diverse requirements. Even within the same sports discipline, each user might have their own specific needs. Often, athletes and coaches have difficulty in articulating their needs clearly and precisely. Professor Hu is actively cultivating cross-disciplinary communication skills among students in the computer science field, aiming to deeply understand the pain points and needs within the sports domain. By crafting specialized AI algorithms and human-computer interaction systems, her team helps users to gain insightful knowledge about themselves and their competitors, thereby securing success across all competitions.
Media Contact：
Ting-Shiuan Chang
Program Manager/Assistant Research Fellow
Department of Engineering and Technologies
National Science and Technology Council
Tel: +886 (2)27377437
