Research map / Detection and segmentation / CNN features
SEGIC: Unleashing the Emergent Correspondence for In-Context Segmentation
Lingchen Meng, Shiyi Lan, Hengduo Li, Jose M Alvarez, Zuxuan Wu, Yu-Gang Jiang
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 geometric, visual and meta instructions together instead of any one alone (ablation) | SegIC (DINOv2-B) · COCO-20i / FSS-1000 / DAVIS-17 | mIoU / J&F: COCO-20i 48.9 (none) -> 74.6 (all three) | Improved |
| use only meta instructions (ablation) | SegIC (DINOv2-B) · FSS-1000 / DAVIS-17 | mIoU / J&F: FSS-1000 63.7, DAVIS-17 21.6 (vs 87.4, 68.4 with all) | Hurt |
| add context reversion and negative entity augmentation together (ablation) | SegIC (DINOv2-B) · COCO-20i / FSS-1000 / DAVIS-17 | mIoU / J&F: COCO-20i 70.0 -> 74.6; DAVIS-17 66.3 -> 68.4 | Improved |
| add either training strategy alone (ablation) | SegIC (DINOv2-B) · DAVIS-17 | J&F: 66.3 -> 65.0 (negative) / 64.3 (reversion) | Hurt |
| add COCO instance segmentation annotations to training (ablation) | SegIC (DINOv2-B) · DAVIS-17 | J&F: 45.1 -> 70.1 | Improved |
| add ADE20k to COCO semantic training data (ablation) | SegIC (DINOv2-B) · FSS-1000 / DAVIS-17 | mIoU / J&F: FSS-1000 80.9 -> 82.3; DAVIS-17 45.1 -> 40.1 | Mixed |
| joint training on COCO-excluded datasets (ADE20k, LVIS, FSS-1000) | SegIC · COCO-20i one-shot | mean mIoU: 53.6 -> 62.3 | Improved |
| use frozen backbone lacking high/multi-resolution support (CLIP-ViT-B, MAE-B) (ablation) | SegIC · COCO-20i | mIoU: 74.6 (DINOv2-B) -> 15.3 (CLIP-ViT-B), 12.0 (MAE-B) | Hurt |
| scale the frozen DINOv2 backbone from B to G (ablation) | SegIC · COCO-20i / FSS-1000 / DAVIS-17 | AVG: 76.8 -> 77.8 | No clear effect |
| use DINOv1 (image self-distillation) instead of DINOv2 backbone (ablation) | SegIC · COCO-20i | mIoU: 74.6 -> 53.5 | Hurt |
| use SegIC (DINOv2-G) vs Matcher on LVIS-92i | SegIC · LVIS-92i | mean mIoU: 33.0 -> 47.8 | Improved |
Later papers that compared with this method
2 later accepted papers report SegIC as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- INSID3: Training-Free In-Context Segmentation with DINOv3 (CVPR 2026 · better on 2 of 4 shared results)
- SANSA: Unleashing the Hidden Semantics in SAM2 for Few-Shot Segmentation (NeurIPS 2025 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Explore In-Context Segmentation via Latent Diffusion Models (AAAI 2025 · similarity 0.88)
- A Simple Image Segmentation Framework via In-Context Examples (NeurIPS 2024 · similarity 0.88)
- Seg-ReSearch: Segmentation with Interleaved Reasoning and External Search (ICML 2026 · similarity 0.87)
- SegMAN: Omni-scale Context Modeling with State Space Models and Local Attention for Semantic Segmentation (CVPR 2025 · similarity 0.87)
- OMG-Seg: Is One Model Good Enough For All Segmentation? (CVPR 2024 · similarity 0.85)
- Bridge the Points: Graph-based Few-shot Segment Anything Semantically (NeurIPS 2024 · similarity 0.85)
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.
