Research map / Label-efficient and robust learning / Domain adaptation and generalization
Zero-Shot Day-Night Domain Adaptation With a Physics Prior
Attila Lengyel, Sourav Garg, Michael Milford, Jan C. van Gemert
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 W color invariant input layer (CIConv) | ResNet-18 · CODaN night test | accuracy (%): 48.31 -> 59.67 | Improved |
| add W color invariant input layer | ResNet-18 · CODaN day test | accuracy (%): 80.39 -> 81.49 | Improved |
| use invariant C instead of RGB | ResNet-18 · CODaN night test | accuracy (%): 48.31 -> 53.44 | Improved |
| use invariant H instead of RGB | ResNet-18 · CODaN day test | accuracy (%): 80.39 -> 75.20 | Hurt |
| add sample-wise normalization to the baseline (ablation) | ResNet-18 · CODaN night test | accuracy (%): 48.31 -> 42.15 | Hurt |
| add log scaling and normalization to the baseline (ablation) | ResNet-18 · CODaN night test | accuracy (%): 48.31 -> 41.90 | Hurt |
| add color augmentation to the baseline (ablation) | ResNet-18 · CODaN night test | accuracy (%): 36.00 -> 48.31 | Improved |
| add color augmentation to W (ablation) | ResNet-18 · CODaN night test | accuracy (%): 53.62 -> 59.67 | Improved |
| replace test-time per-batch statistics (AdaBN) by W | ResNet-18 · CODaN night test | accuracy (%): 55.55 -> 59.67 | Improved |
| learn sigma instead of fixing it (ablation) | W-ResNet-18 · CODaN | accuracy (%) | Improved |
| add W color invariant layer to RefineNet | RefineNet · Nighttime Driving | mIoU: 34.1 -> 41.6 | Improved |
| add W color invariant layer to RefineNet | RefineNet · Dark Zurich | mIoU: 30.6 -> 34.5 | Improved |
Closest papers by meaning
- Similarity Min-Max: Zero-Shot Day-Night Domain Adaptation (ICCV 2023 · similarity 0.92)
- NightCC: Nighttime Color Constancy via Adaptive Channel Masking (CVPR 2024 · similarity 0.87)
- Exploring the Common Appearance-Boundary Adaptation for Nighttime Optical Flow (ICLR 2024 · similarity 0.87)
- Boosting Object Detection with Zero-Shot Day-Night Domain Adaptation (CVPR 2024 · similarity 0.87)
- NightRain: Nighttime Video Deraining via Adaptive-Rain-Removal and Adaptive-Correction (AAAI 2024 · similarity 0.86)
- NightAdapter: Learning a Frequency Adapter for Generalizable Night-time Scene Segmentation (CVPR 2025 · similarity 0.85)
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.
