Research map / Detection and segmentation / 3D detection and driving
3D Adversarial Augmentations for Robust Out-of-Domain Predictions
Alexander Lehner, Stefano Gasperini, Alvaro Marcos-Ramiro, Michael Schmidt, Nassir Navab, Benjamin Busam, Federico Tombari
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 untargeted car adversarial augmentation | Cylinder3D · SemanticKITTI to Waymo | mIoU: 29.43 -> 37.74 | Improved |
| add untargeted car adversarial augmentation | Cylinder3D · SemanticKITTI to nuScenes | mIoU: 29.97 -> 31.55 | Improved |
| add untargeted car adversarial augmentation | Cylinder3D · SemanticKITTI val (in-domain) | mIoU: 64.39 -> 63.85 | Hurt |
| use sample-specific iterative gradient L2 attack as augmentation | Cylinder3D · SemanticKITTI to Waymo | mIoU: 29.43 -> 24.86 | Hurt |
| use sample-specific Chamfer attack as augmentation | Cylinder3D · SemanticKITTI to nuScenes | mIoU: 29.97 -> 22.40 | Hurt |
| add targeted (car to bicycle) augmentation | Cylinder3D · SemanticKITTI val bicycle | IoU: 47.15 -> 54.27 | Improved |
| add untargeted person adversarial augmentation | Cylinder3D · SemanticKITTI to Waymo person | IoU: 6.69 -> 22.22 | Improved |
| add adversarial augmentation (ours) to PointPillars | PointPillars · KITTI to Waymo | AP: 40.86 -> 44.61 | Improved |
| add adversarial augmentation (ours) to Part-A2 | Part-A2 · KITTI to Waymo | AP: 49.76 -> 56.08 | Improved |
| add our augmentation on top of SN domain adaptation | PointPillars + SN · KITTI to Waymo | AP: 49.27 -> 51.32 | Improved |
| add adversarial augmentation (ours) on rare crash cars | PointPillars · KITTI to CrashD rare crash | AP: 22.48 -> 30.37 | Improved |
| add adversarial removal augmentation | PointPillars · KITTI to Waymo | AP: 40.86 -> 40.32 | No clear effect |
Closest papers by meaning
- 3D-VField: Adversarial Augmentation of Point Clouds for Domain Generalization in 3D Object Detection (CVPR 2022 · similarity 0.89)
- Generalizing to Unseen Domains via Adversarial Data Augmentation (NeurIPS 2018 · similarity 0.89)
- An Adversarial Perturbation Oriented Domain Adaptation Approach for Semantic Segmentation (AAAI 2020 · similarity 0.88)
- Evaluating the Adversarial Robustness of Semantic Segmentation: Trying Harder Pays Off (ECCV 2024 · similarity 0.88)
- Adversarial Style Augmentation for Domain Generalized Urban-Scene Segmentation (NeurIPS 2022 · similarity 0.88)
- Density-guided Translator Boosts Synthetic-to-Real Unsupervised Domain Adaptive Segmentation of 3D Point Clouds (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.
