Metric Depth Estimation on Argoverse 2
89.6Delta-1 AccuracyVLM3-4b
Evaluation Results
| Method | Links | |
|---|---|---|
| VLM3-4bModel Type=VLM Trained on Metric Depth Estimation2026.05 | 89.6 | |
| DepthLM-7bModel Type=VLM Trained on Metric Depth Estimation2026.05 | 83.3 | |
| DepthLM-3bModel Type=VLM Trained on Metric Depth Estimation2026.05 | 80.8 | |
| Gemini-2.5-ProModel Type=VLM2026.05 | 28 | |
| GPT-5Model Type=VLM2026.05 | 21.8 | |
| Qwen2.5-VL-72bModel Type=VLM2026.05 | 11.9 | |
| SpaceLLaVA-13BModel Type=Spatial VLM2026.05 | 10 | |
| SpatialRGPT-8BModel Type=Spatial VLM2026.05 | 5.5 | |
| Seed1.5-VLModel Type=VLM Trained on Metric Depth Estimation2026.05 | 4 | |
| Qwen3-VL-32bModel Type=VLM2026.05 | 1.7 | |
| Qwen3-VL-4bModel Type=VLM2026.05 | 0.4 |