Research map / Detection and segmentation / CNN features
VPGNet: Vanishing Point Guided Network for Lane and Road Marking Detection and Recognition
Seokju Lee, Junsik Kim, Jae Shin Yoon, Seunghak Shin, Oleksandr Bailo, Namil Kim, Tae-Hee Lee, Hyun Seok Hong, Seung-Hoon Han, In So Kweon
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add vanishing point prediction task (ablation) | 3-Task network · No-VP set (images without vanishing point) | lane detection F1: 0.4535 -> 0.5234 | Improved |
| train FCN-8s with grid-level instead of pixel-level annotation (ablation) | FCN-8s · No-VP set | lane detection F1: 0.3310 -> 0.4496 | Improved |
| add vanishing point prediction task (ablation) | 3-Task network, single white lane · Scenarios 1-4 | detection+classification F1: Scn1 0.55 -> 0.49; Scn2 0.45 -> 0.52; Scn3 0.31 -> 0.42; Scn4 0.27 -> 0.42 | Mixed |
| add vanishing point prediction task (ablation) | 3-Task network, double yellow lane · Scenarios 1-4 | detection+classification F1: Scn1 0.32 -> 0.36; Scn2 0.62 -> 0.61; Scn3 0.37 -> 0.40; Scn4 0.36 -> 0.40 | Mixed |
| add vanishing point prediction task (ablation) | 3-Task network, dashed white lane · Scenarios 1-4 | detection+classification F1: Scn1 0.77 -> 0.76; Scn2 0.67 -> 0.66; Scn3 0.72 -> 0.73; Scn4 0.68 -> 0.69 | No clear effect |
| add vanishing point prediction task (ablation) | 3-Task network, single yellow lane · Scenarios 1-4 | detection+classification F1: Scn1 0.57 -> 0.58; Scn2 0.64 -> 0.65; Scn3 0.70 -> 0.71; Scn4 0.48 -> 0.42 | Mixed |
| add vanishing point prediction task (ablation) | 3-Task network, straight arrow · Scenarios 1-4 | recall: Scn1 0.46 -> 0.80; Scn2 0.41 -> 0.65; Scn3 0.39 -> 0.63; Scn4 0.48 -> 0.68 | Improved |
| add vanishing point prediction task (ablation) | 3-Task network, crosswalk · Scenarios 1-4 | recall: Scn1 0.88 -> 0.94; Scn2 0.81 -> 0.85; Scn3 0.84 -> 0.93; Scn4 0.82 -> 0.89 | Improved |
| add vanishing point prediction task (ablation) | 3-Task network, safety zone · Scenarios 1-4 | recall: Scn1 0.59 -> 0.80; Scn2 0.47 -> 0.65; Scn3 0.47 -> 0.61; Scn4 0.37 -> 0.38 | Improved |
| add vanishing point prediction task (ablation) | 3-Task network, stop line · Scenarios 1-4 | recall: Scn1 0.83 -> 0.78; Scn2 0.60 -> 0.73; Scn3 0.33 -> 0.56; Scn4 0.60 -> 0.80 | Mixed |
| train VPP-only first phase before joint training (ablation) | VPGNet · VPGNet benchmark | other task losses: about 20% | Improved |
| use grid-level annotation multi-task VPGNet instead of FCN-8s | VPGNet · VPGNet benchmark | forward pass time (ms): 130 -> 30 | Improved |
Closest papers by meaning
- Sketch and Refine: Towards Fast and Accurate Lane Detection (AAAI 2024 · similarity 0.86)
- GEONet: Global Enhancement and Optimization Network for Lane Detection (AAAI 2025 · similarity 0.85)
- ReManNet: A Riemannian Manifold Network for Monocular 3D Lane Detection (CVPR 2026 · similarity 0.85)
- GSENet:Global Semantic Enhancement Network for Lane Detection (AAAI 2024 · similarity 0.85)
- A Hybrid Global-Local Perception Network for Lane Detection (AAAI 2024 · similarity 0.84)
- LaneCPP: Continuous 3D Lane Detection using Physical Priors (CVPR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
