반응형
Abstract
1. Introduction
2. Related Work
2.1. Segmentation-based methods
2.2. Anchor-based methods
2.3. Parameter-based methods
3. Approach
3.1. The Lane Representation
Lane Prior.
3.2. Cross Layer Refinement
Motivation
Refinement structure.
3.3. ROIGather
Motivation
ROIGather structure
3.4. Line IoU loss
Motivation
Formula
3.5. Training and Infercence Details
Positive samples selection.
Training Loss.
4. Experiment
4.1. Datasets
4.2. Implementation details
4.3. Evaluation Metric
4.4. Comparison with the state-of-the-art results
Performance on LLAMAS.
Performance on Tusimple.
4.5. Ablation study
Overall Ablation Study.
Analysis for ROIGather.
Ablation study on Cross Layer Refinement.
Ablation Study on Line-IoU Loss.
5. Conclusion
반응형
댓글