Research map / Detection and segmentation / Open-set and few-shot detection
Increasing the Robustness of Semantic Segmentation Models with Painting-by-Numbers
Christoph Kamann, Carsten Rother
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 Painting-by-Numbers augmentation | DeepLabv3+ Xception-41 · Cityscapes-C Gaussian noise | mIoU: 27.9 -> 46.9 | Improved |
| add Painting-by-Numbers augmentation | DeepLabv3+ ResNet-50 · Cityscapes-C shot noise | mIoU: 12.1 -> 36.1 | Improved |
| add Painting-by-Numbers augmentation | DeepLabv3+ ResNet-50 · Cityscapes-C contrast | mIoU: 52.7 -> 64.2 | Improved |
| add Painting-by-Numbers augmentation | DeepLabv3+ Xception-41 · Cityscapes-C JPEG | mIoU: 38.5 -> 15.7 | Hurt |
| add Painting-by-Numbers augmentation | DeepLabv3+ ResNet-50 · Cityscapes val clean | mIoU: 76.6 -> 76.1 | No clear effect |
| add Painting-by-Numbers augmentation | DeepLabv3+ Xception-41 · Cityscapes-C spatter | mIoU: 46.6 -> 55.8 | Improved |
| add Painting-by-Numbers augmentation | DeepLabv3+ ResNet-50 · Cityscapes-C Gaussian blur | mIoU: 56.5 -> 55.1 | No clear effect |
| add Painting-by-Numbers augmentation | DeepLabv3+ ResNet-50 · Cityscapes-C snow | mIoU: 12.9 -> 10.6 | No clear effect |
| add Painting-by-Numbers augmentation | DeepLabv3+ ResNet-50 · car class with RGB-mean texture | sensitivity (%): 24.5 -> 78.6 | Improved |
| add Painting-by-Numbers augmentation | DeepLabv3+ ResNet-50 · person class under severe noise (scale=1.0) | sensitivity (%): 4.4 -> 52.7 | Improved |
| add Painting-by-Numbers augmentation | DeepLabv3+ ResNet-50 · building class under severe noise (scale=1.0) | sensitivity (%): 94.0 -> 91.3 | Hurt |
| use mean/median class RGB instead of random colors in painting (ablation) | Painting-by-Numbers · Cityscapes-C | robustness | Hurt |
Closest papers by meaning
- Benchmarking the Robustness of Semantic Segmentation Models (CVPR 2020 · similarity 0.91)
- Benchmarking the Robustness of Semantic Segmentation Models with Respect to Common Corruptions (IJCV 2020 · similarity 0.90)
- Evaluating the Adversarial Robustness of Semantic Segmentation: Trying Harder Pays Off (ECCV 2024 · similarity 0.89)
- Reliability in Semantic Segmentation: Are We on the Right Track? (CVPR 2023 · similarity 0.88)
- DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs (TPAMI 2017 · similarity 0.88)
- Towards Reliable Evaluation and Fast Training of Robust Semantic Segmentation Models (ECCV 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.
