Research map / Detection and segmentation / Open-set and few-shot detection
Open-World Semantic Segmentation via Contrasting and Clustering Vision-Language Embedding
Quande Liu, Youpeng Wen, Jianhua Han, Chunjing Xu, Hang Xu, Xiaodan Liang
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 vision-based contrasting (Lvision) (ablation) | ViL-Seg · PASCAL VOC (all classes unseen) | mIoU: 22.05 -> 33.61 | Improved |
| add vision-based contrasting (Lvision) (ablation) | ViL-Seg · PASCAL VOC (all classes unseen) | pixel accuracy: 50.76 -> 75.97 | Improved |
| add vision-based contrasting (Lvision) (ablation) | ViL-Seg · PASCAL Context (all classes unseen) | mIoU: 13.14 -> 15.89 | Improved |
| add vision-based contrasting (Lvision) (ablation) | ViL-Seg · COCO Stuff (all classes unseen) | mIoU: 13.52 -> 16.41 | Improved |
| use online clustering head instead of offline K-means (ablation) | ViL-Seg · PASCAL VOC (all classes unseen) | mIoU: 30.97 -> 33.61 | Improved |
| use online clustering head instead of offline K-means (ablation) | ViL-Seg · PASCAL VOC (all classes unseen) | pixel accuracy: 69.88 -> 75.97 | Improved |
| use online clustering head instead of offline K-means (ablation) | ViL-Seg · PASCAL Context (all classes unseen) | mIoU: 14.82 -> 15.89 | Improved |
| use online clustering head instead of offline K-means (ablation) | ViL-Seg · COCO Stuff (all classes unseen) | mIoU: 15.81 -> 16.41 | Improved |
| use online clustering head instead of offline K-means (ablation) | ViL-Seg · inference speed | cases per second: 8.5 -> 9.8 | Improved |
| use online clustering head instead of offline K-means (ablation) | ViL-Seg · parameters | params (M): 86.19M -> 86.27M | No clear effect |
| vary cluster number C (middle values 20, 25 vs 5, 10, 15 or 30) (ablation) | ViL-Seg · PASCAL VOC, Context, COCO Stuff | mIoU | Mixed |
| add self-training on top of ViL-Seg | ViL-Seg · PASCAL VOC (unseen classes) | mIoU: 34.42 -> 37.30 | Improved |
Closest papers by meaning
- Exploring Open-Vocabulary Semantic Segmentation from CLIP Vision Encoder Distillation Only (ICCV 2023 · similarity 0.92)
- OpenWorldSAM: Extending SAM2 for Universal Image Segmentation with Language Prompts (NeurIPS 2025 · similarity 0.91)
- USE: Universal Segment Embeddings for Open-Vocabulary Image Segmentation (CVPR 2024 · similarity 0.91)
- Training-free Open-Vocabulary Semantic Segmentation via Diverse Prototype Construction and Sub-region Matching (AAAI 2025 · similarity 0.90)
- Emergent Open-Vocabulary Semantic Segmentation from Off-the-shelf Vision-Language Models (CVPR 2024 · similarity 0.90)
- Training-Free Open-Vocabulary Segmentation with Offline Diffusion-Augmented Prototype Generation (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.
