Research map / Label-efficient and robust learning / Domain adaptation and generalization
CMDA: Cross-Modality Domain Adaptation for Nighttime Semantic Segmentation
Ruihao Xia, Chaoqiang Zhao, Meng Zheng, Ziyan Wu, Qiyu Sun, Yang Tang
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 event modality at training time and fuse image+event at test | CMDA · DSEC Night-Semantic | mIoU: 56.0 -> 60.1 vs DAFormer | Improved |
| train with events but test with images only | CMDA · DSEC Night-Semantic | mIoU: 55.1 -> 57.8 | Improved |
| add Image Content-Extractor (ablation) | DAFormer baseline · DSEC Night-Semantic | MIoU(I): 55.06 -> 56.78 | Improved |
| add events via Image Motion-Extractor alone, without random fusion with I_CE (ablation) | DAFormer baseline · DSEC Night-Semantic | MIoU(I): 55.06 -> 53.46 | Hurt |
| combine I_CE and E_ME with random fusion (ablation) | CMDA · DSEC Night-Semantic | MIoU(I+E): 55.65 -> 60.05 | Improved |
| add day-to-night style transfer network (MGCDA) (ablation) | DAFormer baseline · Dark Zurich | mIoU: 53.5 -> 50.2 | Hurt |
| add Image Content-Extractor instead of style transfer | DAFormer baseline · Dark Zurich | mIoU: 53.5 -> 54.8 | Improved |
| use Image Motion-Extractor events instead of ESIM events (ablation) | CMDA · DSEC Night-Semantic | MIoU(E): 42.09 -> 44.91 | Improved |
| add event style transfer network G to E_ME (ablation) | CMDA · DSEC Night-Semantic | MIoU(I+E): 56.63 -> 60.05 | Improved |
| increase shift gamma from 1 to 4 (ablation) | Image Content-Extractor · DSEC Night-Semantic | MIoU(I+E): 60.05 -> 58.57 | Hurt |
| set beta to 0.03 rather than 0.005 (ablation) | Image Content-Extractor · DSEC Night-Semantic | MIoU(I+E): 60.05 -> 57.61 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DAFormer — DAFormer: Improving Network Architectures and Training Strategies for Domain-Adaptive Semantic Segmentation (CVPR 2022) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- A One-Stage Domain Adaptation Network With Image Alignment for Unsupervised Nighttime Semantic Segmentation (TPAMI 2021 · similarity 0.91)
- The Parables of the Mustard Seed and the Yeast: Extremely Low-Budget, High-Performance Nighttime Semantic Segmentation (AAAI 2025 · similarity 0.90)
- NightAdapter: Learning a Frequency Adapter for Generalizable Night-time Scene Segmentation (CVPR 2025 · similarity 0.86)
- Exploring the Common Appearance-Boundary Adaptation for Nighttime Optical Flow (ICLR 2024 · similarity 0.85)
- Efficient Event-Based Semantic Segmentation via Exploiting Frame-Event Fusion: A Hybrid Neural Network Approach (AAAI 2025 · similarity 0.85)
- OpenESS: Event-based Semantic Scene Understanding with Open Vocabularies (CVPR 2024 · 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.
