Injury Prediction(FIELDVISION)
Protecting athletes' knees in 0.05 seconds
With just a single camera and no wearable sensors,
FIELDVISION precisely tracks 34 joint points on an athlete's body in 3D.
"Contactless, precise AI analysis
without wearable sensors"
Without the hassle of attaching sensors to the body,
FIELDVISION analyzes athletes' movements using only a single RGB camera and high-performance edge AI.
Capturing at 60 frames per second, it detects subtle joint movements
and warns in real time of injury risks that are hard to spot with the naked eye.
All 34 joint points, precisely tracked
Using the BodyPose3DNet algorithm, FIELDVISION precisely tracks 34 joint keypoints on the body in 3D. By scientifically analyzing knee angle, trunk tilt, foot spacing, and more, it clearly classifies the level of injury risk.
Lag-free feedback in 0.05 seconds
FIELDVISION uses on-device edge computing that does not rely on the cloud. Optimized parallel GPU processing delivers response times of 50 ms (0.05 seconds) or less, enabling instant posture correction and feedback right on the training ground.
Field-tested with professional teams
Through proof-of-concept (PoC) projects with elite teams, including the Woori Card professional volleyball club, we analyzed more than 1.8 million frames of data and validated the system's performance. The solution is optimized for analyzing landing posture and movement across many sports, including tennis, volleyball, soccer, basketball, and track and field.
