Research map / 3D vision / Human pose estimation
Learning to Predict 3D Lane Shape and Camera Pose from a Single Image via Geometry Constraints
Ruijin Liu, Dapeng Chen, Tie Liu, Zhiliang Xiong, Zejian Yuan
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 |
|---|---|---|---|
| predict camera pose and use it for top-view lane detection with geometry constraints | CLGo · 3D Lane Synthetic Dataset BS, ROS, and SVV | F-Score: 88.1 -> 91.9; 78.0 -> 86.1; 85.3 -> 87.3 | Improved |
| predict camera pose and use it for top-view lane detection with geometry constraints | CLGo · FLCP | F1: 30.87 -> 33.82 | Improved |
| add the auxiliary 3D lane branch to Stage 1 (ablation) | CLGo Stage 1 · 3D Lane Synthetic Dataset SVV | F-Score: 85.5 -> 87.3 | Improved |
| add Stage 1 reprojection geometry constraints (ablation) | CLGo Stage 1 · 3D Lane Synthetic Dataset SVV | F-Score: 85.5 -> 87.3 | Improved |
| add flat-ground constraint to Stage 2 (ablation) | CLGo Stage 2 · 3D Lane Synthetic Dataset SVV | F-Score: 87.3 -> 87.1 | No clear effect |
| add image-plane constraint to Stage 2 (ablation) | CLGo Stage 2 · 3D Lane Synthetic Dataset SVV | F-Score: 87.1 -> 85.3 | Hurt |
| train the two stages jointly rather than separately (ablation) | CLGo · 3D Lane Synthetic Dataset SVV | F-Score: 84.9 -> 87.3 | Improved |
| use ground-truth lanes instead of Stage 1 estimated lanes for geometry constraints (ablation) | CLGo · 3D Lane Synthetic Dataset SVV | F-Score: 87.3 -> 86.3 | Hurt |
| use Stage 2 lane estimates instead of Stage 1 estimates for geometry constraints (ablation) | CLGo · 3D Lane Synthetic Dataset SVV | F-Score: 87.3 -> 85.4 | Hurt |
| replace transformer with fully connected layers (ablation) | CLGo · 3D Lane Synthetic Dataset SVV | F-Score: 87.3 -> 81.1 | Hurt |
| share features between the two stages instead of applying IPM to images (ablation) | CLGo · 3D Lane Synthetic Dataset SVV | F-Score: Image 87.3; Feat.C 87.2; Feat.M 86.7; Feat.R4 86.8; Memory 86.5 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Gen-LaneNet — Gen-LaneNet: A Generalized and Scalable Approach for 3D Lane Detection (ECCV 2020) · better on 7 of 7 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Rethinking Lanes and Points in Complex Scenarios for Monocular 3D Lane Detection (CVPR 2025 · similarity 0.88)
- LaneCPP: Continuous 3D Lane Detection using Physical Priors (CVPR 2024 · similarity 0.88)
- PVALane: Prior-Guided 3D Lane Detection with View-Agnostic Feature Alignment (AAAI 2024 · similarity 0.88)
- DV-3DLane: End-to-end Multi-modal 3D Lane Detection with Dual-view Representation (ICLR 2024 · similarity 0.87)
- GLane3D: Detecting Lanes with Graph of 3D Keypoints (CVPR 2025 · similarity 0.87)
- Sketch and Refine: Towards Fast and Accurate Lane Detection (AAAI 2024 · similarity 0.86)
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.
