Research map / Label-efficient and robust learning / OOD and anomaly detection
RoboDepth: Robust Out-of-Distribution Depth Estimation under Corruptions
Lingdong Kong, Shaoyuan Xie, Hanjiang Hu, Lai Xing Ng, Benoit R Cottereau, Wei Tsang Ooi
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 |
|---|---|---|---|
| Use a Transformer-based MDE model rather than a conventional CNN | MDE models · KITTI-C | mCE: 100.00 -> 74.95 | Improved |
| Train using monocular images rather than stereo or mixed monocular and stereo inputs (ablation) | MonoDepth2 · KITTI-C | mCE: 100.00 -> 124.31 | Improved |
| Add ImageNet pre-training (ablation) | MonoDepth2 · KITTI-C | robustness to weather, lighting, and data-processing corruptions | Mixed |
| Train with higher-resolution inputs (ablation) | MDE models · KITTI-C noise corruptions | robustness: 30% | Improved |
| Increase the number of trainable parameters | MDE models · corruption robustness | robustness | Hurt |
| Increase training complexity | MDE models · corruption robustness | robustness | Improved |
| Fine-tune on the same corruption used for evaluation (ablation) | MonoDepth2 · KITTI-C | robustness to that corruption | Improved |
| Fine-tune using severe weather or movement corruptions (ablation) | MonoDepth2 · KITTI-C | overall robustness | Hurt |
| Apply simulated corruption fine-tuning (ablation) | MonoDepth2 · nuScenes and Foggy-Cityscapes | distribution gap | Improved |
| Evaluate on stylized images with texture cues diminished | MDE models · KITTI-S | mDEE | Hurt |
| Use a model with global awareness | MDE models · KITTI-S | mDEE | Improved |
Closest papers by meaning
- DepthVanish: Optimizing Adversarial Interval Structures for Stereo-Depth-Invisible Patches (NeurIPS 2025 · similarity 0.87)
- Synthetic-to-Real Self-supervised Robust Depth Estimation via Learning with Motion and Structure Priors (CVPR 2025 · similarity 0.87)
- Evaluating Robustness of Monocular Depth Estimation with Procedural Scene Perturbations (NeurIPS 2025 · similarity 0.87)
- SharpDepth: Sharpening Metric Depth Predictions Using Diffusion Distillation (CVPR 2025 · similarity 0.87)
- Adaptive Fusion of Single-View and Multi-View Depth for Autonomous Driving (CVPR 2024 · similarity 0.87)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · 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.
