1 paper
Tiancheng Zhang, Mengmeng Wang, Yan Gao +3
Video lane detection requires predictions that remain stable across frames, yet severe vehicle occlusions can break temporal cues. In streaming recurrent models, corrupted observat…