Research map / Label-efficient and robust learning / Domain adaptation and generalization
Exploiting Domain Properties in Language-Driven Domain Generalization for Semantic Segmentation
Seogkyu Jeon, Kibeom Hong, Hyeran Byun
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 texture perturbation (ablation) | DPMFormer (CLIP ViT-B) · GTAV to unseen (ViT-B), average | mIoU: 54.97 -> 55.38 | Improved |
| add consistency loss L_cons (ablation) | DPMFormer (CLIP ViT-B) · GTAV to unseen (ViT-B), average | mIoU: 55.38 -> 56.15 | Improved |
| add domain-aware contrastive prompt loss L_contra (ablation) | DPMFormer (CLIP ViT-B) · GTAV to unseen (ViT-B), average | mIoU: 56.15 -> 58.14 | Improved |
| add texture perturbation (Cityscapes target only) (ablation) | DPMFormer (CLIP ViT-B) · GTAV to Cityscapes | mIoU: 57.5 -> 57.04 | Hurt |
| replace L_contra with CoCoOp contrastive loss (ablation) | DPMFormer (CLIP ViT-B) · GTAV to unseen (ViT-B), average | mIoU: 56.13 -> 56.36 | No clear effect |
| CoCoOp+ with augmented sample as positive (ablation) | DPMFormer (CLIP ViT-B) · GTAV to unseen (ViT-B), average | mIoU: 56.13 -> 56.20 | No clear effect |
| replace L_contra with PromptSRC (ablation) | DPMFormer (CLIP ViT-B) · GTAV to unseen (ViT-B), average | mIoU: 56.13 -> 56.78 | Improved |
| use similarity on prompt embeddings sim(pi,pi) instead of sim(t,t) (ablation) | L_contra · GTAV to unseen (ViT-B), average | mIoU: 56.57 -> 58.14 | Improved |
| use sim(pi,pi) instead of sim(t,v) (ablation) | L_contra · GTAV to unseen (ViT-B), average | mIoU: 57.34 -> 58.14 | Improved |
| lambda_cons 10 versus 1 (ablation) | DPMFormer (CLIP ViT-B) · GTAV to unseen (ViT-B), average | mIoU: 55.80 -> 58.14 | Improved |
| lambda_cons 10 versus 50 (ablation) | DPMFormer (CLIP ViT-B) · GTAV to unseen (ViT-B), average | mIoU: 56.98 -> 58.14 | Improved |
| temperature tau 0.5 versus 2 (ablation) | DPMFormer (CLIP ViT-B) · GTAV to unseen (ViT-B), average | mIoU: 55.72 -> 58.14 | Improved |
Closest papers by meaning
- Exploring Semantic Consistency and Style Diversity for Domain Generalized Semantic Segmentation (AAAI 2025 · similarity 0.92)
- MaskViM: Domain Generalized Semantic Segmentation with State Space Models (AAAI 2025 · similarity 0.90)
- Collaborating Foundation Models for Domain Generalized Semantic Segmentation (CVPR 2024 · similarity 0.90)
- A Simple Recipe for Language-guided Domain Generalized Segmentation (CVPR 2024 · similarity 0.90)
- Domain Generalization in CLIP via Learning with Diverse Text Prompts (CVPR 2025 · similarity 0.90)
- Style Blind Domain Generalized Semantic Segmentation via Covariance Alignment and Semantic Consistence Contrastive Learning (CVPR 2024 · similarity 0.90)
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.
