Problem
Research questions and friction points this paper is trying to address.
Video Lane Detection
Temporal Consistency
Vehicle Occlusion
State Contamination
State Space Model
SIGMA-Lane: Scale-pyramId Gated MAmba for Temporally Consistent Video Lane Detection
🤖 AI Summary
This study addresses temporal state contamination and prediction instability caused by severe occlusion in video lane detection. We propose SIGMA-Lane, which models occlusion-induced failures as state pollution and incorporates an occlusion-aware dual-gating mechanism within the State Space Model pathway. By integrating coordinate-consistent affine alignment with a structural spatial retrieval module, the method effectively filters noise and restores lane topology. Experiments on the VIL-100 and OpenLane-V datasets demonstrate that SIGMA-Lane significantly enhances temporal stability under heavy occlusion, achieving state-of-the-art F1-score and mIoU performance. These results validate the model’s robustness in complex dynamic environments, offering a reliable solution for maintaining consistent lane predictions despite challenging visual conditions.
📝 Abstract
Video lane detection requires predictions that remain stable across frames, yet severe vehicle occlusions can break temporal cues. In streaming recurrent models, corrupted observations may enter the hidden state and produce errors that persist into later frames. Existing occlusion-aware refinements usually provide obstacle masks as auxiliary inputs, so the state-update path is only indirectly protected. We propose SIGMA-Lane, which treats this failure mode as state contamination in State Space Model (SSM)-based temporal modeling. SIGMA-Lane places occlusion-aware gates on the SSM write and residual-fusion paths, controlling how current observations enter temporal memory and are fused back after temporal propagation. After coordinate-consistent affine alignment, the model combines two complementary paths: SSM-consistent dual-gating for temporal filtering and Structural Spatial Retrieval (SSR) for recovering missing lane structure from aligned historical priors. Experiments on VIL-100 and OpenLane-V show improved temporal stability under heavy occlusion, with competitive F1 and mIoU scores.
Innovation
Methods, ideas, or system contributions that make the work stand out.
State Space Model
Occlusion-aware Gating
Temporal Consistency
Structural Spatial Retrieval
Video Lane Detection
🔎 Similar Papers
No similar papers found.
