Research map / Label-efficient and robust learning / Domain adaptation and generalization
Project to Adapt: Domain Adaptation for Depth Completion from Noisy and Sparse Sensor Data
Adrian Lopez-Rodriguez, Benjamin Busam, Krystian Mikolajczyk
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 real LiDAR mask sampling instead of Bernoulli sparsification (ablation) | synthetic-only FusionNet · KITTI depth completion (validation, ablation) | RMSE: 1975.14 -> 1608.32 | Improved |
| add projection artifacts on top of LiDAR mask (ablation) | synthetic-only FusionNet · KITTI depth completion (validation, ablation) | RMSE: 1608.32 -> 1335.00 | Improved |
| add projection to Bernoulli sampled input (ablation) | synthetic-only FusionNet · KITTI depth completion (validation, ablation) | RMSE: 1975.14 -> 3286.73 | Hurt |
| add CycleGAN RGB adaptation (ablation) | synthetic-only FusionNet · KITTI depth completion (validation, ablation) | RMSE: 1335.00 -> 1247.53 | Improved |
| add reliable-point filtering S_p with MAE loss on real data (ablation) | two-step pipeline · KITTI depth completion (validation, ablation) | RMSE: 1315.74 -> 1150.27 | Improved |
| use MAE instead of BerHu for real-data supervision with S_p (ablation) | two-step pipeline · KITTI depth completion (validation, ablation) | RMSE: 1328.76 -> 1150.27 | Improved |
| add RGB input (ablation) | FusionNet sparse-depth model · KITTI depth completion (validation, ablation) | RMSE: 1175.54 -> 1167.83 | No clear effect |
| use CycleGAN-adapted RGB instead of raw RGB (ablation) | FusionNet · KITTI depth completion (validation, ablation) | RMSE: 1167.83 -> 1150.27 | Improved |
| use prior image transfer method instead of CycleGAN (ablation) | FusionNet · KITTI depth completion (validation, ablation) | RMSE: 1167.83 -> 1184.39 | Hurt |
| add output discriminator to DA Base (ablation) | DA Base · KITTI depth completion (validation) | RMSE: 1630.31 -> 1636.89 | No clear effect |
| add feature discriminator to DA Base (ablation) | DA Base · KITTI depth completion (validation) | RMSE: 1630.31 -> 1617.41 | No clear effect |
| add video self-supervision photometric loss | Ours · KITTI depth completion (validation) | RMSE: 1150.27 -> 1112.83 | Improved |
Closest papers by meaning
- Flexible Depth Completion for Sparse and Varying Point Densities (CVPR 2024 · similarity 0.92)
- Distilling Monocular Foundation Model for Fine-grained Depth Completion (CVPR 2025 · similarity 0.89)
- Test-Time Adaptation for Depth Completion (CVPR 2024 · similarity 0.89)
- Zero-shot Depth Completion via Test-time Alignment with Affine-invariant Depth Prior (AAAI 2025 · similarity 0.87)
- Masked Spatial Propagation Network for Sparsity-Adaptive Depth Refinement (CVPR 2024 · similarity 0.87)
- Mind The Edge: Refining Depth Edges in Sparsely-Supervised Monocular Depth Estimation (CVPR 2024 · 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.
