Research map / 3D vision / Depth and stereo
Improving Domain Generalization in Self-Supervised Monocular Depth Estimation via Stabilized Adversarial Training
Yuanqi Yao, Gang Wu, Kui Jiang, Siao Liu, Jian Kuai, Xianming Liu, Junjun Jiang
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 SCAT (SDN + CGS adversarial training) to MonoDepth2 | MonoDepth2 · KITTI-C | mCE (%): 101.04 -> 86.32 | Improved |
| add SCAT to MonoViT | MonoViT · KITTI-C | mCE (%): 80.54 -> 62.74 | Improved |
| add SCAT to Robust-Depth | Robust-Depth · KITTI-C | mCE (%): 55.72 -> 53.37 | Improved |
| use vanilla adversarial data augmentation instead of none | MonoViT · KITTI Eigen test (clean) | Abs Rel: 0.099 -> 0.106 | Hurt |
| use SCAT rather than vanilla ADA on clean images | MonoViT · KITTI Eigen test (clean) | Abs Rel: 0.106 -> 0.100 | Improved |
| add SCAT to MonoDepth2 on night driving | MonoDepth2 · NuScenes-Night | Abs Rel: 0.397 -> 0.355 | Improved |
| add offline data augmentation (DA) instead of adversarial augmentation | MonoDepth2 · KITTI-C | mCE (%): 101.04 -> 97.72 | Improved |
| add Conflict Gradient Surgery alone to adversarial training (ablation) | MonoDepth2 ADA · KITTI-C | Abs Rel: 0.193 -> 0.174 | Improved |
| add Scaling Depth Network alone to adversarial training (ablation) | MonoDepth2 ADA · KITTI-C | Abs Rel: 0.193 -> 0.179 | Improved |
| combine CGS and SDN (ablation) | MonoDepth2 ADA · KITTI-C | Abs Rel: 0.193 -> 0.165 | Improved |
| increase adversarial perturbation size beyond 135 (ablation) | SCAT MonoDepth2 · KITTI-C | Abs Rel: 0.165 -> 0.177 | Hurt |
| use small perturbation size 20 instead of 135 (ablation) | SCAT MonoDepth2 · KITTI-C | Abs Rel: 0.165 -> 0.196 | Hurt |
Closest papers by meaning
- Self-Supervised Adversarial Training of Monocular Depth Estimation Against Physical-World Attacks (TPAMI 2024 · similarity 0.91)
- Adversarial Training of Self-supervised Monocular Depth Estimation against Physical-World Attacks (ICLR 2023 · similarity 0.91)
- Semi-Supervised Adversarial Monocular Depth Estimation (TPAMI 2019 · similarity 0.90)
- Self-Supervised Deep Monocular Depth Estimation With Ambiguity Boosting (TPAMI 2021 · similarity 0.88)
- Deep Digging into the Generalization of Self-Supervised Monocular Depth Estimation (AAAI 2023 · similarity 0.88)
- Digging Into Self-Supervised Monocular Depth Estimation (ICCV 2019 · 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.
