AI-Sports '21: 2nd International Workshop on Artificial Intelligence in Sports

In conjunction with the IEEE International Conference on Multimedia and Expo (ICME) 2021

Organizers

Huang-Chia Shih, Rainer Lienhart, Takahiro Ogawa, Thomas B. Moeslund, Jenq-Neng Hwang

Descriptions

Utilization of various sensors for bioinformatics data acquisition has become increasingly popular in recent years. Meanwhile, research fields like computer vision, sensing technology, wearable technology, machine learning and data-driven approaches recently made huge advancements and have massively impacted many aspects of sports, the joint assessment of multiple modalities for sport data analytics offers appealing innovations to advance the field. Data-driven machine learning technique plays an important role in developing and improving sports in recent years. Coaches and athletes are able to utilize this data to make better decisions for developing their team. Popular sports like football fuel the drive for technological advances in AI and machine learning. With the current technology, specific details and strategies can be extracted from the data to help coaches and players see the whole picture with clarity. By adding context to the collected data, coaches and analysts can allocate more time towards developing strategies. The growing number of potential programs shows promising technological advances in the sports industry, but limits are becoming more obvious. There is not enough data to effectively create learning and artificial intelligence. Sporting research has already gone underway by big companies like Google and Facebook. It is apparent that big data sports analytics is a strong, positive correlation to maximizing a sport teamˇ¦s potential, and any team will find themselves at a severe disadvantage if they do not actively incorporate themselves with big data analytics.

Scope and Topics

This workshop is open to anyone interested in sports content analytics. To cover the rapid progress of emerging areas we plan to focus our target field in three topics:

• Sports tactic analysis using machine learning and visual processing. This session aims at applications in domains such as event reasoning and tactic analysis. In offline service, historical records can be used to analyze video content through machine learning. In online service, discovered latent knowledge can be used for real-time tactic recommendation. Recently, optimization of player positioning, posture, and movement with deep learning method is attracting much attention.
• Applying big data/machine learning techniques to personalized athletic training and rehabilitation. This session highlights integration of computation into athletic training and recovery. There are tons of data on sports and health care being collected but very few of them are analyzed. There is enormous potential in the data to revolutionize the sports industry and to drastically improve athletesˇ¦ performance and health. Big data analytics in sports are uncommon but history has revealed that utilizing the data correlates with faster and efficient improvement in an athlete'sˇ¦ performance.
• Data-driven ghosting and prediction for sports. This workshop plans to publish papers presenting novel ghosting and prediction schemes. Sports data is analyzed to create ˇ§ghostsˇ¨ of players and enables them to visually study a situation and compare what the ghosts would have done and what they should have done. The research field includes human computer interaction, AR/VR/MR data visualization, and UX/UI interaction design with human factors.

Date and time

Friday, July 9, 2021 14:00 UTC+8.

Schedule

Session Chairs: Huang-Chia Shih (Yuan Ze University, Taiwan) and Takahiro Ogawa (Hokkaido University, Japan)

Start	End	Paper	Authors
14:00	14:05	Opening
14:05	14:35	Keynote: Pose Analytics in Real-world Problem Solving	Phokgoan Chioh (CTO, IdeaLab.com, USA)
14:35	14:50	Self-supervised Learning for Human Pose Estimation in Sports	Katja Ludwig*, Sebastian Scherer, Moritz Einfalt, and Rainer Lienhart (University of Augsburg, Germany)
14:50	15:05	Spatiotemporal-Spectral Graph Convolutional Networks for Skeleton-Based Action Recognition	Shuo Chen*, Xinghao Jiang, Tanfeng Sun, and Ke Xu (Shanghai Jiao Tong University, China)
15:05	15:20	Swimmer Stroke Rate Estimation From Overhead Race Video	Timothy Woinoski* and Ivan Bajic (Simon Fraser University, Canada)
15:20	15:35	Center of Mass Trajectory: An Image Descriptor for Baseball Swing Analysis Based on A Single Low Cost Camera	Chih-Chieh Fang1, Ching-Hsien Hsu1, Chun-Wen Chiu2, Jung-Tang Kung2, and Huang-Chia Shih* 1 1(Yuan Ze University, Taiwan), 2 (National Taiwan Sport University, Taiwan).
15:35	15:40	Closing remarks

Previous workshops:

1st AI-Sports20' in conjunction with ICME 2020