Research map / Detection and segmentation / Open-set and few-shot detection
Exemplar-FreeSOLO: Enhancing Unsupervised Instance Segmentation With Exemplars
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 exemplar knowledge abstraction and exemplar embedding contrastive guidance | FreeSOLO · MS COCO val2017 instance segmentation | AP: 4.0 -> 8.4 | Helped |
| add exemplar knowledge abstraction and exemplar embedding contrastive guidance | FreeSOLO object detection · MS COCO val2017 | AP: 5.5 -> 12.6 | Helped |
| add exemplar knowledge abstraction and exemplar embedding contrastive guidance | FreeSOLO · UVO val instance segmentation | AP: 4.8 -> 9.2 | Helped |
| add exemplar knowledge abstraction and exemplar embedding contrastive guidance | FreeSOLO · PASCAL VOC trainval07 multi-object discovery | AP: 10.2 -> 12.6 | Helped |
| replace the unsupervised exemplar mechanism with box ground-truth foreground supervision (ablation) | Exemplar-FreeSOLO · MS COCO val2017 instance segmentation | AP: 8.4 -> 5.8 | Hurt |
| replace the unsupervised exemplar mechanism with mask ground-truth foreground supervision (ablation) | Exemplar-FreeSOLO · MS COCO val2017 instance segmentation | AP: 8.4 -> 6.2 | Hurt |
| replace the unsupervised exemplar mechanism with box ground-truth foreground supervision (ablation) | Exemplar-FreeSOLO object detection · MS COCO val2017 | AP: 12.6 -> 7.1 | Hurt |
| replace the unsupervised exemplar mechanism with mask ground-truth foreground supervision (ablation) | Exemplar-FreeSOLO object detection · MS COCO val2017 | AP: 12.6 -> 8.2 | Hurt |
| increase the number of exemplar objects per category from 1 to 9 (ablation) | Exemplar-FreeSOLO · MS COCO val2017 instance segmentation | AP: 8.4 -> 8.9 | Helped |
| increase the number of exemplar objects per category from 1 to 9 (ablation) | Exemplar-FreeSOLO object detection · MS COCO val2017 | AP: 12.6 -> 13.0 | Helped |
| vary the exemplar embedding contrastive loss weight (ablation) | Exemplar-FreeSOLO · MS COCO val2017 instance segmentation | AP: 7.6, 7.6, 8.8, 8.8, 8.2, 8.2, 8.4, 8.1 | Mixed |
| vary the exemplar embedding contrastive loss weight (ablation) | Exemplar-FreeSOLO object detection · MS COCO val2017 | AP: 10.8, 10.7, 11.1, 11.9, 12.0, 12.2, 12.6, 11.8 | Mixed |
Closest papers by meaning
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · similarity 0.86)
- Training-Free Dataset Pruning for Instance Segmentation (ICLR 2025 · similarity 0.86)
- unMORE: Unsupervised Multi-Object Segmentation via Center-Boundary Reasoning (ICML 2025 · similarity 0.86)
- Unsupervised Universal Image Segmentation (CVPR 2024 · similarity 0.85)
- Segment Anything without Supervision (NeurIPS 2024 · similarity 0.85)
- EAGLE: Eigen Aggregation Learning for Object-Centric Unsupervised Semantic Segmentation (CVPR 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.