3D Object Detection on nuScenes (val) (AP by Instance Count) benchmark leaderboard · SOTA2 Research

3D Object Detection on nuScenes (val) (AP by Instance Count)

54.6mAP (All)FOMO-3D

Evaluation Results

Method	Links
FOMO-3D2026.04		54.6	79.9	59.6	27.6
MMLF2026.04		51.4	77.9	59.4	20
BEVFusion2026.04		45.5	75.5	52	12.8
CMT2026.04		44.4	79.9	53	4.8
MMF2026.04		43.6	77.1	49	9.4
BEVFusion-L2026.04		42.5	72.5	48	10.6
SemLT3D2026.04		41.1	62.08	40.62	20.77
CenterPoint2026.04		40.4	77.1	45.1	4.3
TransFusion2026.04		39.8	73.9	41.2	9.8
CenterPoint2026.04		39.2	76.4	43.1	3.5
SemLT3D2026.04		29.59	50.92	34.55	6.03
SparseBEV2026.04		27.94	54.18	29.67	4.1
RayDN2026.04		27.41	54.5	31.02	0.6
StreamPETR2026.04		26.97	53.32	28.53	3.22
PETRv22026.04		18.13	38.82	14.96	0.3
BEVDet4D2026.04		17.98	39.34	18.07	0
BEVFormer2026.04		15.51	34.6	14.91	0.3
BEVDepth2026.04		15.43	36.46	13.51	0.13
BEVDet2026.04		13.27	27.7	14.31	0