Research map / Label-efficient and robust learning / Test-time adaptation
Continual Adaptation for Deep Stereo
Matteo Poggi, Alessio Tonioni, Fabio Tosi, Stefano Mattoccia, Luigi Di Stefano
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 online adaptation (FULL, photometric loss) | MADNet · KITTI City | D1-all (%): 37.42 -> 3.35 | Improved |
| use SGM proxy supervision instead of photometric self-supervision (MAD) | MAD · KITTI City | D1-all (%): 7.51 -> 4.12 | Improved |
| use SGM proxy supervision instead of photometric self-supervision (MAD) | MAD · KITTI Campus | D1-all (%): 22.27 -> 11.24 | Improved |
| use SGM proxy supervision instead of photometric loss (FULL) | FULL · KITTI Campus | D1-all (%): 10.62 -> 9.69 | Improved |
| use SGM proxy instead of photometric loss (FULL) | FULL · KITTI City | D1-all (%): 3.35 -> 3.51 | No clear effect |
| use WILD proxy instead of photometric loss (FULL) | FULL · KITTI City | D1-all (%): 3.35 -> 5.11 | Hurt |
| use LIDAR proxy instead of photometric loss (MAD) | MAD · KITTI Campus | D1-all (%): 22.27 -> 19.33 | Improved |
| use proxy-supervised FULL++ instead of FULL | MADNet-Synthia · across domains | D1-all (%): 2.76 -> 2.23 | Improved |
| add FOL2A meta-learning to FULL | MADNet-Synthia · across domains | D1-all (%): 2.76 -> 2.60 | Improved |
| use MAD (modular) instead of FULL adaptation | MADNet · KITTI across domains | D1-all (%): 2.43 -> 4.09 | Hurt |
| add prior adaptation on KITTI before DrivingStereo (FULL) | MADNet · DrivingStereo Dusky | D1-all (%): 20.40 -> 9.21 | Improved |
| add prior adaptation on KITTI before DrivingStereo (MAD) | MADNet · DrivingStereo Dusky | D1-all (%): 33.23 -> 9.81 | Improved |
Closest papers by meaning
- Real-Time Self-Adaptive Deep Stereo (CVPR 2019 · similarity 0.96)
- Unsupervised Adaptation for Deep Stereo (ICCV 2017 · similarity 0.92)
- Reusable Architecture Growth for Continual Stereo Matching (TPAMI 2024 · similarity 0.92)
- RobIA: Robust Instance-aware Continual Test-time Adaptation for Deep Stereo (NeurIPS 2025 · similarity 0.90)
- Federated Online Adaptation for Deep Stereo (CVPR 2024 · similarity 0.89)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.88)
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.
