3D Object Detection on Lyft (val)
40.62AP (Car)BEVFusion + VERIA (InternVL3/MoGe2)
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| BEVFusion + VERIA (InternVL3/MoGe2)Modality=Multi-modal, Base Detector=BEVFusion, Augmentation=VERIA, VLM Semantic Verifier=InternVL3, Depth Estimator=MoGe22026.03 | 40.62 | 17.33 | 30.35 | 4.9 | 5.85 | 18.54 | 17.99 | 3.06 | |
| VERIA (InternVL3/MoGe2)Modality=Multi-modal, Base detector=BEVFusion, VLM=InternVL3, Depth Estimator=MoGe22026.03 | 40.62 | 17.33 | 30.35 | 4.9 | 5.85 | 18.54 | 17.99 | 3.06 | |
| BEVFusion + VERIA (Qwen3VL/UniDepth2)Modality=Multi-modal, Base Detector=BEVFusion, Augmentation=VERIA, VLM Semantic Verifier=Qwen3VL, Depth Estimator=UniDepth22026.03 | 40.52 | 17.18 | 30.23 | 4.95 | 6.22 | 18.7 | 17.15 | 2.47 | |
| VERIA (Qwen3VL/UniDepth2)Modality=Multi-modal, Base detector=BEVFusion, VLM=Qwen3VL, Depth Estimator=UniDepth22026.03 | 40.52 | 17.18 | 30.23 | 4.95 | 6.22 | 18.7 | 17.15 | 2.47 | |
| BEVFusionModality=Multi-modal, Base Detector=BEVFusion2026.03 | 40.51 | 16.97 | 30.34 | 4.98 | 4.85 | 18.81 | 17.72 | 1.6 | |
| BEVFusionModality=Multi-modal2026.03 | 40.51 | 16.97 | 30.34 | 4.98 | 4.85 | 18.81 | 17.72 | 1.6 | |
| BEVFusion + VERIA (Qwen3VL/MoGe2)Modality=Multi-modal, Base Detector=BEVFusion, Augmentation=VERIA, VLM Semantic Verifier=Qwen3VL, Depth Estimator=MoGe22026.03 | 40.39 | 17.25 | 30.88 | 5.01 | 6.17 | 18.71 | 17.66 | 1.94 | |
| VERIA (Qwen3VL/MoGe2)Modality=Multi-modal, Base detector=BEVFusion, VLM=Qwen3VL, Depth Estimator=MoGe22026.03 | 40.39 | 17.25 | 30.88 | 5.01 | 6.17 | 18.71 | 17.66 | 1.94 | |
| BEVFusion + VERIA (InternVL3/UniDepth2)Modality=Multi-modal, Base Detector=BEVFusion, Augmentation=VERIA, VLM Semantic Verifier=InternVL3, Depth Estimator=UniDepth22026.03 | 40.12 | 17.14 | 30.56 | 5.15 | 5.51 | 18.63 | 17.66 | 2.37 | |
| VERIA (InternVL3/UniDepth2)Modality=Multi-modal, Base detector=BEVFusion, VLM=InternVL3, Depth Estimator=UniDepth22026.03 | 40.12 | 17.14 | 30.56 | 5.15 | 5.51 | 18.63 | 17.66 | 2.37 | |
| CenterPoint + VERIA (InternVL3/MoGe2)Modality=LiDAR only, Base Detector=CenterPoint, Augmentation=VERIA, VLM Semantic Verifier=InternVL3, Depth Estimator=MoGe22026.03 | 36.69 | 17.92 | 30.73 | 6.02 | 6.02 | 19.17 | 21.27 | 5.52 | |
| CenterPoint + VERIA (InternVL3/UniDepth2)Modality=LiDAR only, Base Detector=CenterPoint, Augmentation=VERIA, VLM Semantic Verifier=InternVL3, Depth Estimator=UniDepth22026.03 | 36.69 | 17.94 | 31.12 | 6.03 | 6.26 | 19.47 | 20.86 | 5.12 | |
| VERIA (InternVL3/MoGe2)Modality=LiDAR only, Base detector=CenterPoint, VLM=InternVL3, Depth Estimator=MoGe22026.03 | 36.69 | 17.92 | 30.73 | 6.02 | 6.02 | 19.17 | 21.27 | 5.52 | |
| VERIA (InternVL3/UniDepth2)Modality=LiDAR only, Base detector=CenterPoint, VLM=InternVL3, Depth Estimator=UniDepth22026.03 | 36.69 | 17.94 | 31.12 | 6.03 | 6.26 | 19.47 | 20.86 | 5.12 | |
| CenterPoint + VERIA (Qwen3VL/MoGe2)Modality=LiDAR only, Base Detector=CenterPoint, Augmentation=VERIA, VLM Semantic Verifier=Qwen3VL, Depth Estimator=MoGe22026.03 | 36.6 | 17.98 | 30.8 | 6.1 | 5.9 | 20.7 | 21.13 | 4.6 | |
| VERIA (Qwen3VL/MoGe2)Modality=LiDAR only, Base detector=CenterPoint, VLM=Qwen3VL, Depth Estimator=MoGe22026.03 | 36.6 | 17.98 | 30.8 | 6.1 | 5.9 | 20.7 | 21.13 | 4.6 | |
| CenterPoint + GT-AugModality=LiDAR only, Base Detector=CenterPoint, Augmentation=GT-Aug2026.03 | 36.4 | 17.37 | 30.6 | 5.8 | 5.1 | 18.6 | 20.7 | 4.4 | |
| CenterPoint + GT-AugModality=LiDAR only, Base detector=CenterPoint2026.03 | 36.4 | 17.37 | 30.6 | 5.8 | 5.1 | 18.6 | 20.7 | 4.4 | |
| CenterPoint + Text3DAugModality=LiDAR only, Base Detector=CenterPoint, Augmentation=Text3DAug2026.03 | 36.2 | 17.7 | 30.7 | 6.1 | 5.4 | 19.1 | 21.9 | 4.5 | |
| CenterPoint + Text3DAugModality=LiDAR only, Base detector=CenterPoint2026.03 | 36.2 | 17.7 | 30.7 | 6.1 | 5.4 | 19.1 | 21.9 | 4.5 | |
| CenterPoint + VERIA (Qwen3VL/UniDepth2)Modality=LiDAR only, Base Detector=CenterPoint, Augmentation=VERIA, VLM Semantic Verifier=Qwen3VL, Depth Estimator=UniDepth22026.03 | 36.1 | 17.71 | 30.9 | 6.1 | 6.2 | 19.44 | 20.36 | 4.9 | |
| VERIA (Qwen3VL/UniDepth2)Modality=LiDAR only, Base detector=CenterPoint, VLM=Qwen3VL, Depth Estimator=UniDepth22026.03 | 36.1 | 17.71 | 30.9 | 6.1 | 6.2 | 19.44 | 20.36 | 4.9 |