Object Counting on Cross-domain General Object Counting v1 (test)
2.08MAE (Sheep)REC-RL
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| REC-RLModel size=7B2026.05 | 2.08 | 46.15 | 11.12 | 12.01 | 8.63 | 6.45 | 14.4 | |
| REC-RLModel size=3B2026.05 | 3.05 | 83.75 | 18.74 | 14.31 | 9.34 | 6.64 | 22.64 | |
| Q-7B-FRModel size=7B, Training=FR2026.05 | 3.33 | 88.1 | 20.6 | 14.52 | 9.71 | 8.16 | 24.07 | |
| LLaMA-90BModel size=90B2026.05 | 3.87 | 182.4 | 28.6 | 18.18 | 19.24 | 5.7 | 44.68 | |
| Q-3B-SFTModel size=3B, Training=SFT2026.05 | 4.19 | 104.34 | 33.38 | 18.57 | 10.68 | 8.15 | 29.89 | |
| Q-3B-FRModel size=3B, Training=FR2026.05 | 4.19 | 85.52 | 24.02 | 18.86 | 10.55 | 7.81 | 25.16 | |
| Q-7B-SFTModel size=7B, Training=SFT2026.05 | 4.86 | 97.41 | 24.78 | 18.51 | 11.63 | 7.93 | 27.52 | |
| Q-3B-R1Model size=3B, Training=R12026.05 | 5.75 | 97.43 | 41.74 | 25.99 | 14.78 | 8.92 | 32.44 | |
| GPT-4o2026.05 | 7.12 | 243.41 | 58.36 | 39.18 | 17.22 | 5.82 | 61.95 | |
| Q-7B-R1Model size=7B, Training=R12026.05 | 8.04 | 145.55 | 32.29 | 22.68 | 14.42 | 9.88 | 38.81 | |
| In-VL2-26BModel size=26B2026.05 | 14.13 | 271.96 | 64.19 | 33.64 | 22.4 | 8.22 | 69.09 |