Queensland uses AI to screen potential Olympians

The Queensland government has developed an AI-powered application on Azure to screen budding Olympic athletes using computer vision.

Launched by the Queensland Academy of Sport, the custom-built YouFor2032 app features a series of virtual tests to assess prospective athletes’ strength, speed, power, endurance, balance and flexibility.

Tests are assessed using video-based markerless motion capture and pose estimation to generate key points from a user’s self-recordings of exercise tasks.

The AI maps joint movements and extracts patterns to analyse performance, measuring metrics such as jump height, fingertip reach, and the number of correct push-up repetitions.

The data captured by the app is used to determine which athletes are shortlisted for face-to-face assessments with QAS staff.

“The AI and computer vision provides the QAS talent team with quantitative data to develop a holistic view of the athlete’s current abilities,” a spokesperson for the state’s Department of Tourism and Sport told Digital Nation.

“This data is received by the QAS talent team, who use the test results along with age-matched norms, athlete background and interest in specific sports to create an athlete profile.

“These profiles are then matched to the requirements of YouFor2032 partner sports, and athletes who are close to the thresholds are invited to face-to-face testing with the QAS talent team.”

From a technical perspective, the app uses React Native for its mobile framework and Django on the backend, while TensorFlow powers the AI components.