| Paper | Code | J&F | ModelName | ReleaseDate |
|---|---|---|---|---|
| SAM 3: Segment Anything with Concepts | ✓ Link | 84.4 | SAM 3 (zero-shot) (extra data) | 2025-11-20 |
| SAM2Long: Enhancing SAM 2 for Long Video Segmentation with a Training-Free Memory Tree | ✓ Link | 81.2 | SAM2.1Long-L (extra data) | 2024-10-21 |
| SAM 2: Segment Anything in Images and Videos | ✓ Link | 78.4 | SAM 2 (Hiera-L) (extra data) | 2024-08-01 |
| Efficient Track Anything | ✓ Link | 74.5 | EfficientTAM-S (extra data) | 2024-11-28 |
| EdgeTAM: On-Device Track Anything Model | ✓ Link | 71.7 | EdgeTAM (extra data) | 2025-01-13 |
| SAM-I2V++: Efficiently Upgrading SAM for Promptable Video Segmentation | ✓ Link | 67.8 | SAM-I2V++ | 2025-12-26 |
| MobileSAM2: Lightweight Segment Anything in Images and Videos via Hypergraphical Knowledge Distillation | 67.8 | MobileSAM2-23M | 2026-09-08 | |
| SAM-I2V: Upgrading SAM to Support Promptable Video Segmentation with Less than 0.2% Training Cost | ✓ Link | 65.2 | TinySAM + SAM-I2V (32-GPU scaling run) (extra data) | 2025-06-02 |