Research map / Detection and segmentation / Open-set and few-shot detection
Large-Scale Interactive Object Segmentation With Human Annotators
Check 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 |
|---|---|---|---|
| Use region-center clicks instead of boundary clicks (ablation) | M_b + M_b+c interactive segmentation · COCO_L test, after three rounds | mIoU: 77% -> 80% | Helped |
| Add 3-pixel Gaussian click noise (ablation) | M_b + M_b+c interactive segmentation · COCO_L test, three rounds | mIoU: 3% drop | Hurt |
| Add 6-pixel Gaussian click noise (ablation) | M_b + M_b+c interactive segmentation · COCO_L test, three rounds | mIoU: 7% drop | Hurt |
| Ignore error regions smaller than 20^2 pixels (ablation) | M_b + M_b+c interactive segmentation · COCO_L test, three rounds | mIoU: 3% | Hurt |
| Ignore error regions smaller than 30^2 pixels (ablation) | M_b + M_b+c interactive segmentation · COCO_L test, three rounds | mIoU: 8% | Hurt |
| Encode clicks as Gaussian maps or distance transforms rather than binary disks (ablation) | M_b+c · COCO_L test | mIoU | Hurt |
| Vary the binary-disk click diameter (ablation) | M_b+c · COCO_L test | mIoU | No clear effect |
| Add the previous-round mask as an input channel (ablation) | M_b+c · COCO_L test | mIoU | No clear effect |
| Distribute about nine clicks over multiple rounds rather than all in one round (ablation) | M_b + M_b+c interactive segmentation · COCO_L test | mIoU | Helped |
| Increase the total number of corrective clicks (ablation) | M_b + M_b+c interactive segmentation · COCO_L test | mIoU | Helped |
| Use class-agnostic models instead of class-specific models (ablation) | M_b and M_b+c · COCO_L test, after three rounds | mIoU: within 2 percent points | No clear effect |
| Train on ADE or COCO while excluding the target car or giraffe category (ablation) | M_b and M_b+c · COCO_L test, after three rounds | mIoU | No clear effect |
Closest papers by meaning
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · similarity 0.88)
- COCONut: Modernizing COCO Segmentation (CVPR 2024 · similarity 0.87)
- TETRIS: Towards Exploring the Robustness of Interactive Segmentation (AAAI 2024 · similarity 0.87)
- Training-Free Dataset Pruning for Instance Segmentation (ICLR 2025 · similarity 0.87)
- AGILE3D: Attention Guided Interactive Multi-object 3D Segmentation (ICLR 2024 · similarity 0.86)
- Any3DIS: Class-Agnostic 3D Instance Segmentation by 2D Mask Tracking (CVPR 2025 · similarity 0.86)
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.