AI analysis system aids assessments of athletes

Special: 2022 Winter Olympics

A team from Fudan University has developed a training analysis system based on artificial intelligence to assist in performance assessments of winter sports athletes taking part in the Beijing 2022 Winter Olympic Games, which will begin on Feb 4.

Zhang Lihua, the team leader and a professor of engineering and technology at the Shanghai university, said the system analyzes athletes' movements, postures, speed and other performance indicators during training through algorithms, contributing to the creation of customized training plans for each athlete and the overall efficiency of scientific training.

He initiated the idea of establishing the system in 2019, when the General Administration of Sport sought to promote the winter sports equipment industry for the Games.

His idea was also supported by Kristin Collins, a sports performance consultant at the administration's Winter Sports Management Center in Beijing.

"It was two years of collaborative work with the sports administration in Jilin province, a northeastern region often covered in snow during winter that is suitable for sports training," Zhang said.

Normally, coaches observe athletes in training to give feedback on how they are performing, but the analysis system enables real-time digital monitoring of the athletes through high-definition cameras, which identify every move they make, and summarizes their performance, Zhang said.

"It takes that burden off coaches and enhances training efficiency through quantitative scientific analysis," he said.

For sports like figure skating, the system is also able to improve the standardization and aesthetics of athletes' performance postures, thanks to its 3D detection and pose estimation features, he added.

Other sports, like soccer, can also use the system for real-time monitoring and analysis, Zhang said.

"The children of a friend in the United States are fond of ice hockey and want to use the system at home to elevate their performance," he said. "The system could possibly be used not merely for serving national sports competitions, but also among the general public."

Zhang said the team has been researching the system's functions for sports competitions that are open to novice and amateur athletes, for the creation of AI auxiliary judges and for the development of intelligent sports venues and facilities.