Visual Counting on VIEW2SPACE v1
0.58MAEOurs (Grounded CoT)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Ours (Grounded CoT)Category=Fine-tuned, Backbone=Qwen3VL-4B2026.03 | 0.58 | 54.99 | |
| Ours (CoT)Category=Fine-tuned, Backbone=Qwen3VL-4B2026.03 | 0.68 | 51.43 | |
| Ours (Instruct-tuning)Category=Fine-tuned, Backbone=Qwen3VL-4B2026.03 | 0.71 | 50.08 | |
| GPT-5-miniCategory=Closed-source MLLMs2026.03 | 1.2 | 38.71 | |
| GPT-5.2Category=Closed-source MLLMs2026.03 | 1.28 | 36.5 | |
| GPT-5Category=Closed-source MLLMs2026.03 | 1.31 | 38.1 | |
| GPT-4oCategory=Closed-source MLLMs2026.03 | 1.51 | 29.03 | |
| GPT-5-nanoCategory=Closed-source MLLMs2026.03 | 1.73 | 24.28 | |
| Internvl2.5-8BCategory=Open-source MLLMs2026.03 | 2.04 | 22.5 | |
| Gemma-3-12B-itCategory=Open-source MLLMs2026.03 | 2.08 | 24.53 | |
| Qwen3-VL-8BCategory=Open-source MLLMs2026.03 | 2.1 | 24.92 | |
| Spatial-MLLMCategory=Open-source Spatial Models2026.03 | 2.1 | 9.98 | |
| DeepSeek-VL2-SmallCategory=Open-source MLLMs2026.03 | 2.2 | 24.37 | |
| Molmo2-4BCategory=Open-source Spatial Models2026.03 | 2.21 | 22.17 | |
| RoBobrain2.0Category=Open-source Spatial Models2026.03 | 2.22 | 21.49 | |
| Qwen3-VL-4BCategory=Open-source MLLMs2026.03 | 2.26 | 21.15 | |
| Random (frequency)Category=Baseline2026.03 | 2.46 | 13.05 | |
| OpenviewCategory=Open-source Spatial Models2026.03 | 2.57 | 19.8 | |
| Qwen3-VL-2BCategory=Open-source MLLMs2026.03 | 2.63 | 16.34 | |
| Qwen2.5-VL-7BCategory=Open-source MLLMs2026.03 | 2.7 | 13.03 | |
| Mantis-8B (SigLip)Category=Open-source MLLMs2026.03 | 2.94 | 17.6 | |
| Idefics2-8BCategory=Open-source MLLMs2026.03 | 3.92 | 11.34 | |
| MINDCUBE-AugCategory=Open-source Spatial Models2026.03 | 4.52 | 18.1 | |
| MINDCUBE-PlainCategory=Open-source Spatial Models2026.03 | 4.77 | 19.12 | |
| Molmo2-8BCategory=Open-source Spatial Models2026.03 | 5.93 | 21.66 | |
| Ours (Grounded CoT, Direct Answer at Test)Category=Fine-tuned, Backbone=Qwen3VL-4B2026.03 | 10.75 | 29.1 | |
| Random (chance)Category=Baseline2026.03 | 11.37 | 3.39 |