FisheyeDistanceNet: Self-Supervised Scale-Aware Distance Estimation using Monocular Fisheye Camera for Autonomous Driving
FisheyeDistanceNet is a self-supervised method that learns Euclidean distance and ego-motion from raw fisheye video sequences for autonomous driving. It uses CNNs to regress a distance map by synthesizing target images from preceding and succeeding frames, overcoming scale ambiguity through velocity estimates, and employing specialized loss functions to ensure accurate metric depth estimation.