MAD
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Video Grounding | MAD (test) | Recall@1 (IoU=0.1)17.3 | 35 | |
| Synthetic in-context reasoning | MAD synthetic (test) | Compression Score55.5 | 24 | |
| Video Grounding | MAD 1.0 (test) | R@1 (IoU=0.1)12.43 | 17 | |
| Movie Audio Description generation | MAD-eval-Named v2 (test) | C Score28.2 | 17 | |
| Human motion segmentation | MAD | Precision (Pr)0.7043 | 14 | |
| Anomaly Segmentation | MAD (test) | AUROC98.4 | 11 | |
| Anomaly Localization | MAD | Gorilla99.5 | 11 | |
| Long Video Moment Retrieval | MAD (test) | Recall@1 (Tol 0.1)15 | 10 | |
| Tabular Classification | MAD M (test) | Macro F1 Score0.861 | 9 | |
| Anomaly Detection | MAD | Gorilla93.6 | 9 | |
| Single-image Morphing Attack Detection | MAD22 | EER (OpenCV)0.3 | 8 | |
| Audio Description Generation | MAD-eval-Named (test) | Originality1.49 | 8 | |
| Anomaly Detection | MAD-sys (test) | AUROC (Image-level)95.6 | 6 | |
| Audio Description Generation | MAD-eval Named | R-L12.1 | 4 | |
| Synthetic token manipulation | MAD benchmark | Compression Score43.8 | 2 | |
| Synthetic in-context reasoning | MAD | Compress- | 0 |
Showing 16 of 16 rows