Danger detection in skiing using a helmet-mounted 360-degree camera

Skiing is a winter sport widely enjoyed around the world, but it carries the risk of accidents due to collisions and falls. The objective of this research is to develop a system that uses a helmet-mounted 360-degree camera to detect other skiers in the vicinity and avoid the risk of collisions. The system utilizes the object detection algorithm YOLOv8 and the skeletal estimator ViTPose to make risk decisions based on the skier’s position, direction, and posture. Experiments were conducted using data collected at an indoor ski resort to verify the accuracy of the system in detecting other skiers and detecting danger.