Research map / Detection and segmentation / Open-set and few-shot detection
Retrieve and Segment: Are a Few Examples Enough to Bridge the Supervision Gap in Open-Vocabulary Segmentation?
Tilemachos Aravanis, Vladan Stojnić, Bill Psomas, Nikos Komodakis, Giorgos Tolias
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 class relevance weights w_c (ablation) | RnS · Avg mIoU 6 benchmarks, B=1 | mIoU: 41.20 -> 41.59 | Improved |
| add class relevance weights w_c (ablation) | RnS · Avg mIoU 6 benchmarks, B=10 | mIoU: 48.54 -> 49.02 | Improved |
| use multiple mixing coefficients Lambda instead of single 0.8 (ablation) | RnS · Avg mIoU 6 benchmarks, B=1 | mIoU: 36.40 -> 41.20 | Improved |
| use multiple mixing coefficients Lambda instead of single 0.8 (ablation) | RnS · Avg mIoU 6 benchmarks, B=10 | mIoU: 48.38 -> 48.54 | Improved |
| add textual support (class names) (ablation) | RnS · Avg mIoU 6 benchmarks, B=1 | mIoU: 34.11 -> 41.59 | Improved |
| add textual support (class names) (ablation) | RnS · Avg mIoU 6 benchmarks, B=10 | mIoU: 48.00 -> 49.02 | Improved |
| use K=4 neighbours instead of K=1 in kNN retrieval (ablation) | RnS · Avg mIoU 6 benchmarks, B=1 | mIoU: 40.02 -> 41.59 | Improved |
| increase K from 4 to 32 (ablation) | RnS · Avg mIoU 6 benchmarks, B=5 | mIoU: 47.87 -> 47.59 | No clear effect |
| add pseudo-label loss for classes without visual support (ablation) | RnS · partial visual support | mIoU | Improved |
| retrieve nearest support features instead of random subset of support set (ablation) | RnS · retrieval analysis | mIoU | Improved |
| use RnS on top of zero-shot DINOv3.txt + SAM at B=20 | DINOv3.txt + SAM · OVS vs fully supervised avg | mIoU: 27.9 -> 61.9 | Improved |
| add visual support (partial, 90% classes unseen) | RnS · VOC all-class mIoU | mIoU: 52.85 -> 52.04 | No clear effect |
Closest papers by meaning
- Emergent Open-Vocabulary Semantic Segmentation from Off-the-shelf Vision-Language Models (CVPR 2024 · similarity 0.93)
- Test-Time Adaptation of Vision-Language Models for Open-Vocabulary Semantic Segmentation (NeurIPS 2025 · similarity 0.91)
- Transferable and Principled Efficiency for Open-Vocabulary Segmentation (CVPR 2024 · similarity 0.91)
- Training-free Open-Vocabulary Semantic Segmentation via Diverse Prototype Construction and Sub-region Matching (AAAI 2025 · similarity 0.91)
- Relationship Prompt Learning is Enough for Open-Vocabulary Semantic Segmentation (NeurIPS 2024 · similarity 0.91)
- OpenWorldSAM: Extending SAM2 for Universal Image Segmentation with Language Prompts (NeurIPS 2025 · similarity 0.91)
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.
