Research map / 3D vision / Depth and stereo
CARD: A Multi-Modal Automotive Dataset for Dense 3D Reconstruction in Challenging Road Topography
Gasser Elazab, Frank Neuhaus, Tilman Koß, Malte Splietker, Aditya Date, Michael Unterreiner, Maximilian Jansen, Olaf Hellwich
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 |
|---|---|---|---|
| fine-tune MoGe2L on CARD with affine-invariant loss plus L1 height-space loss | MoGe2L · CARD full image | AbsRel: 0.045 -> 0.029 | Improved |
| fine-tune MoGe2L on CARD with affine-invariant loss plus L1 height-space loss | MoGe2L · CARD bounding boxes | Height AbsDiff: 0.051 -> 0.034 | Improved |
| fine-tune MoGe2L on CARD with affine-invariant loss plus L1 height-space loss | MoGe2L · CARD full image | Height AbsDiff: 0.119 -> 0.051 | Improved |
| fine-tune UniDepthV2-L decoder on CARD with plain L1 depth loss | UniDepthV2-L · CARD full image | RMSE: 1.825 -> 1.749 | Improved |
| fine-tune UniDepthV2-L decoder on CARD with plain L1 depth loss | UniDepthV2-L · CARD bounding boxes | AbsRel: 0.029 -> 0.027 | No clear effect |
| replace monocular teacher with FoundationStereo and use L1 metric loss in DMD3C | DMD3C · CARD depth completion (full image) | RMSE [m]: 0.7742 -> 0.7510 | Improved |
| replace monocular teacher with FoundationStereo and use L1 metric loss in DMD3C | DMD3C · CARD depth completion (bounding boxes) | delta1: 0.9900 -> 0.9984 | Improved |
| replace monocular teacher with FoundationStereo and use L1 metric loss in DMD3C | DMD3C · CARD depth completion (full image) | iMAE: 0.0050 -> 0.0057 | Hurt |
| use stereo (FoundationStereo) instead of best monocular model | MoGe2L · CARD bounding boxes | Height AbsDiff: 0.051 -> 0.025 | Improved |
| apply per-image median scaling to monocular predictions | DAV2 · CARD full image | AbsRel: 1.537 -> 0.096 | Improved |
| add LiDAR-only depth completion on top of stereo teacher (DMD3C+FS vs FoundationStereo) | DMD3C+FS · CARD bounding boxes | RMSE: 0.1918 -> 0.185 | Hurt |
| add voxel cleaning and baseline-adaptive voting to GT aggregation (ablation) | LiDAR GT pipeline · CARD ground truth (qualitative) | dynamic-point artifacts | Improved |
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.
- DepthPro — Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025) · better on 4 of 4 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
- DrivingStereo: A Large-Scale Dataset for Stereo Matching in Autonomous Driving Scenarios (CVPR 2019 · similarity 0.90)
- The Road Ahead in Autonomous Driving: The KITScenes Multimodal Dataset (NeurIPS 2026 · similarity 0.88)
- Effective Multi-sensor Conditioning for Street-view Novel-view Synthesis (NeurIPS 2026 · similarity 0.87)
- SS3DM: Benchmarking Street-View Surface Reconstruction with a Synthetic 3D Mesh Dataset (NeurIPS 2024 · similarity 0.87)
- ViiNeuS: Volumetric Initialization for Implicit Neural Surface Reconstruction of Urban Scenes with Limited Image Overlap (CVPR 2025 · similarity 0.87)
- Spotting the Unexpected (STU): A 3D LiDAR Dataset for Anomaly Segmentation in Autonomous Driving (CVPR 2025 · similarity 0.87)
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.
