Metric: AP@50 (higher is better)
| # | Model↕ | AP@50▼ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | WeakSAM-Mask2Former (with SAM) | 39.9 | No | WeakSAM: Segment Anything Meets Weakly-supervise... | 2024-02-22 | Code |
| 2 | WeakSAM-Mask RCNN (with SAM) | 34.5 | No | WeakSAM: Segment Anything Meets Weakly-supervise... | 2024-02-22 | Code |
| 3 | BESTIE (proposal-free) | 34.1 | No | Beyond Semantic to Instance Segmentation: Weakly... | 2021-09-20 | Code |
| 4 | CIM + Mask R-CNN | 29.7 | Yes | Complete Instances Mining for Weakly Supervised ... | 2024-02-12 | Code |
| 5 | BESTIE (proposal-free) | 28 | No | Beyond Semantic to Instance Segmentation: Weakly... | 2021-09-20 | Code |
| 6 | LIID | 27.1 | Yes | - | - | Code |
| 7 | Fan et al. | 25.5 | Yes | - | - | - |
| 8 | CIM | 23 | Yes | Complete Instances Mining for Weakly Supervised ... | 2024-02-12 | Code |