Danger Detection for Skiing via a Helmet-Mounted 360-Degree Camera

Abstract

It is difficult for skiers to detect danger when other approaching skiers are in their blind spot. In this paper, we propose an image-based system that detects danger in all directions via a helmet-mounted 360-degree camera. Other skiers are detected using YOLO and the risk is judged based on the distance at which they are present. We also detect danger by using a bird's eye view to estimate skiing direction.