Semantic Segmentation for Lanes in Running Track to Assist Blind Athletes

This work presents a technological solution to increase the autonomy of blind athletes or with visual disability in their training, using neural networks to segment every lane of the track field in real-time along with a device that captures the video signal in front of the athlete and indicates by vibrations when a curve is approaching or when a deviation from the lane is occurring. The results show great performance in the segmentation of the lanes and the timely indications to the athlete to keep him in the lane.