Research map / Label-efficient and robust learning / Domain adaptation and generalization
Label-Free Text Prototype Adaptation for Open Vocabulary Segmentation
Keli Wang, Meixuan Li, Tianyu Li, Guoqing 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 |
|---|---|---|---|
| add prototype refinement (fixed step) (ablation) | SegEarth-OV · Vaihingen | mIoU: 29.1 -> 31.5 | Improved |
| add confident pixel selection (ablation) | +P · Vaihingen | mIoU: 31.5 -> 32.9 | Improved |
| add reliability score and scaled step size (ablation) | +P +A · Vaihingen | mIoU: 32.9 -> 35.3 | Improved |
| add reliability-aware orthogonalisation (ablation) | +P +A +B · Vaihingen | mIoU: 32.9 -> 34.9 | Improved |
| combine scaled step and orthogonalisation (ablation) | +P +A +B · Vaihingen | mIoU: 35.3 -> 37.2 | Improved |
| add pixel selection to reliability estimation (ablation) | VPTA · Vaihingen | mIoU: 35.6 -> 37.2 | Improved |
| use reliability-descending ordering instead of random (ablation) | VPTA · Vaihingen | mIoU: 36.4 -> 37.2 | Improved |
| replace class names with stronger naming (set C) (ablation) | SegEarth-OV · Vaihingen | mIoU: 29.1 -> 40.1 | Improved |
| apply VPTA under stronger naming (ablation) | SegEarth-OV · Vaihingen (C naming) | mIoU: 40.1 -> 41.5 | Improved |
| adapt on LoveDA RGB images, evaluate on Vaihingen (ablation) | VPTA · Vaihingen | mIoU: 40.1 -> 21.9 | Hurt |
| pool Vaihingen and Potsdam images for adaptation (ablation) | VPTA · Vaihingen | mIoU: 40.1 -> 42.0 | Improved |
| use large uniform step beta=0.30 instead of reliability-scaled step (ablation) | VPTA · LoveDA | mIoU | Hurt |
Closest papers by meaning
- Test-Time Adaptation of Vision-Language Models for Open-Vocabulary Semantic Segmentation (NeurIPS 2025 · similarity 0.89)
- Emergent Open-Vocabulary Semantic Segmentation from Off-the-shelf Vision-Language Models (CVPR 2024 · similarity 0.88)
- Training-free Open-Vocabulary Semantic Segmentation via Diverse Prototype Construction and Sub-region Matching (AAAI 2025 · similarity 0.87)
- Toward Modality Gap: Vision Prototype Learning for Weakly-supervised Semantic Segmentation with CLIP (AAAI 2025 · similarity 0.87)
- OpenWorldSAM: Extending SAM2 for Universal Image Segmentation with Language Prompts (NeurIPS 2025 · similarity 0.87)
- Training-Free Open-Vocabulary Segmentation with Offline Diffusion-Augmented Prototype Generation (CVPR 2024 · similarity 0.87)
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.
