Research map / Detection and segmentation / Open-set and few-shot detection
SAN: Side Adapter Network for Open-Vocabulary Semantic Segmentation
Mengde Xu, Zheng Zhang, Fangyun Wei, Han Hu, Xiang Bai
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 CLIP feature fusion (ablation) | SAN · ADE-150 | mIoU: 21.1 -> 27.8 | Improved |
| fuse multiple layers instead of 9th layer only (ablation) | SAN · ADE-150 | mIoU: 27.1 -> 27.8 | Improved |
| fuse deeper (9th) instead of stem layer (ablation) | SAN · ADE-150 | mIoU: 20.0 -> 27.1 | Improved |
| single-forward instead of twice-forward CLIP (ablation) | SAN · ADE-150 | mIoU: 27.6 -> 27.8 | No clear effect |
| 9 fusion and 3 recognition layers instead of 3 and 9 (ablation) | SAN · ADE-150 | mIoU: 23.8 -> 27.8 | Improved |
| end-to-end training instead of two-stage (ablation) | SAN · ADE-150 | mIoU: 21.6 -> 26.1 | Improved |
| decoupled head instead of single head (ablation) | SAN · ADE-150 | mIoU: 26.1 -> 27.8 | Improved |
| low-resolution CLIP input 320 instead of 640 (ablation) | SAN · ADE-150 | mIoU: 24.6 -> 27.8 | Improved |
| fine-tune CLIP position embedding (ablation) | SAN · ADE-150 | mIoU: 27.0 -> 27.8 | Improved |
| increase SAN width from 240 to 288 (ablation) | SAN · ADE-150 | mIoU: 27.8 -> 27.3 | No clear effect |
| per-head attention bias (ablation) | SAN · ADE-150 | mIoU: 26.2 -> 27.8 | Improved |
| add layer-wise attention bias (ablation) | SAN · ADE-150 | mIoU: 27.8 -> 27.4 | Hurt |
Closest papers by meaning
- Effective SAM Combination for Open-Vocabulary Semantic Segmentation (CVPR 2025 · similarity 0.90)
- Mask-Adapter: The Devil is in the Masks for Open-Vocabulary Segmentation (CVPR 2025 · similarity 0.89)
- Emergent Open-Vocabulary Semantic Segmentation from Off-the-shelf Vision-Language Models (CVPR 2024 · similarity 0.89)
- Exploring Simple Open-Vocabulary Semantic Segmentation (CVPR 2025 · similarity 0.88)
- Test-Time Adaptation of Vision-Language Models for Open-Vocabulary Semantic Segmentation (NeurIPS 2025 · similarity 0.88)
- Unveiling the Knowledge of CLIP for Training-Free Open-Vocabulary Semantic Segmentation (AAAI 2025 · 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.
