Research map / Label-efficient and robust learning / Domain adaptation and generalization
Both Style and Distortion Matter: Dual-Path Unsupervised Domain Adaptation for Panoramic Semantic Segmentation
Xu Zheng, Jinjing Zhu, Yexin Liu, Zidong Cao, Chong Fu, Lin Wang
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 |
|---|---|---|---|
| use ERP and tangent images in dual paths instead of ERP in both paths (ablation) | DPPASS-T · DensePASS test | mIoU: 45.22 -> 49.53 | Improved |
| add intra-projection loss L_g (ablation) | DPPASS-T · DensePASS test | mIoU: 42.40 -> 50.12; +7.72 | Improved |
| add prediction consistency loss L_pc (ablation) | DPPASS-T · DensePASS test | mIoU: 42.40 -> 49.53; +7.13% | Improved |
| add tangent-wise feature contrastive loss L_fc (ablation) | DPPASS-T · DensePASS test | mIoU: 42.40 -> 47.30; +4.90% | Improved |
| combine all four losses rather than supervised loss alone (ablation) | DPPASS-T · DensePASS test | mIoU: 42.40 -> 55.30; +12.9 | Improved |
| use tangent projection size 224 × 224 rather than the tested alternatives (ablation) | DPPASS-T · DensePASS test | mIoU: 49.98, 52.22, 55.30, 55.17, 52.56 | Mixed |
| set alpha to 0.02 instead of the other tested values (ablation) | DPPASS-T · DensePASS test | mIoU: 48.25, 50.01, 50.12, 48.12, 47.44 | Mixed |
| set beta to 50 instead of the other tested values (ablation) | DPPASS-T · DensePASS test | mIoU: 46.97, 47.21, 49.53, 49.38, 42.1 | Mixed |
| use DPPASS-S instead of Trans4PASS-S | Segformer-B2 · DensePASS test | mIoU: 55.22 -> 56.28; +1.06% | Improved |
| use DPPASS-T instead of Trans4PASS-T | Segformer-B1 · WildPASS test | mIoU(%): 54.67 -> 60.38; 5.71% | Improved |
| use DPPASS-S instead of Trans4PASS-S | Segformer-B2 · WildPASS test | mIoU(%): 62.91 -> 63.53; 0.62% | Improved |
Closest papers by meaning
- Semantics Distortion and Style Matter: Towards Source-free UDA for Panoramic Segmentation (CVPR 2024 · similarity 0.95)
- GoodSAM: Bridging Domain and Capacity Gaps via Segment Anything Model for Distortion-aware Panoramic Semantic Segmentation (CVPR 2024 · similarity 0.92)
- AdapToPASS: Ambiguity-aware Adaptive Spherical Transformer for Panoramic Semantic Segmentation (NeurIPS 2026 · similarity 0.87)
- PanoWorld: Towards Spatial Supersensing in 360◦ Panorama World (NeurIPS 2026 · similarity 0.87)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.86)
- Open-Set Domain Adaptation for Semantic Segmentation (CVPR 2024 · similarity 0.86)
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.
