Research map / 3D vision / 3D generation
A Parametric Top-View Representation of Complex Road Scenes
Ziyan Wang, Buyu Liu, Samuel Schulter, Manmohan Chandraker
Check 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 |
|---|---|---|---|
| Use hybrid training with real and simulated BEV data instead of simulated data only | H-BEV+DA vs S-BEV+DA · KITTI | IOU: .314 -> .398 | Improved |
| Use hybrid training with real and simulated BEV data instead of simulated data only | H-BEV+DA vs S-BEV+DA · NuScenes | IOU: .171 -> .346 | Improved |
| Add domain adaptation to simulated-only BEV training (ablation) | S-BEV · KITTI | Accu.-Mc.: .371 -> .677 | Improved |
| Add domain adaptation to simulated-only BEV training (ablation) | S-BEV · NuScenes | Accu.-Bi.: .790 -> .753 | Hurt |
| Add depth prediction supervision (ablation) | M-RGB · KITTI | MSE: .230 -> .146 | Improved |
| Use BEV input instead of RGB input | M-BEV vs M-RGB+D · KITTI | IOU: .342 -> .345 | Improved |
| Use BEV input instead of RGB input | M-BEV vs M-RGB+D · NuScenes | IOU: .335 -> .269 | Hurt |
| Add domain adaptation to hybrid training (ablation) | H-BEV · KITTI | MSE: .152 -> .108 | Improved |
| Add domain adaptation to hybrid training (ablation) | H-BEV · NuScenes | Accu.-Mc.: .569 -> .545 | Hurt |
| Use separate domain-specific feature extractor weights without a discriminator (ablation) | H-BEV · KITTI | IOU: .342 -> .381 | Improved |
| Add a graphical model (ablation) | H-BEV+DA · KITTI | IOU: .398 -> .371 | Hurt |
| Add a graphical model (ablation) | H-BEV+DA · NuScenes | IOU: .346 -> .450 | Improved |
Closest papers by meaning
- Learning to Look around Objects for Top-View Representations of Outdoor Scenes (ECCV 2018 · similarity 0.91)
- Understanding Road Layout From Videos as a Whole (CVPR 2020 · similarity 0.90)
- RoadSceneBench: A Lightweight Benchmark for Mid-Level Road Scene Understanding (CVPR 2026 · similarity 0.86)
- Multi-Level Neural Scene Graphs for Dynamic Urban Environments (CVPR 2024 · similarity 0.85)
- Dynamic 3D Gaussian Fields for Urban Areas (NeurIPS 2024 · similarity 0.85)
- ViiNeuS: Volumetric Initialization for Implicit Neural Surface Reconstruction of Urban Scenes with Limited Image Overlap (CVPR 2025 · similarity 0.85)
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.
