3D Object Detection on Waymo Open Dataset (val)
75.313D APH Vehicle L2MSF
Evaluation results
| Method | Links | |||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MSFFrames=82023.03 | 75.31 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.65 | — | — | — | — | — | — | — | — | 75.46 | 82.83 | 82.01 | 75.76 | 85.24 | 82.21 | 78.32 | 75.61 | 78.52 | 77.74 | 76.32 | 75.47 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MPPNetFrames=162023.03 | 74.96 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.4 | — | — | — | — | — | — | — | — | 74.85 | 82.74 | 82.28 | 75.41 | 84.69 | 82.25 | 77.43 | 75.06 | 77.28 | 76.66 | 75.13 | 74.52 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CenterFormerFrames=82023.03 | 73.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.3 | — | — | — | — | — | — | — | — | 73.7 | 78.8 | 78.3 | 74.3 | 82.1 | 79.3 | 77.8 | 75 | 75.2 | 74.4 | 73.2 | 72.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MPPNetFrames=42023.03 | 73.61 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.83 | — | — | — | — | — | — | — | — | 74.22 | 81.54 | 81.06 | 74.07 | 84.56 | 81.94 | 77.2 | 74.67 | 77.15 | 76.5 | 75.01 | 74.38 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MSFFrames=42023.03 | 73.35 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.2 | — | — | — | — | — | — | — | — | 74.62 | 81.36 | 80.87 | 73.81 | 85.05 | 82.1 | 77.92 | 75.11 | 78.4 | 77.61 | 76.17 | 75.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CenterFormerFrames=42023.03 | 72.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77 | — | — | — | — | — | — | — | — | 73.2 | 78.1 | 77.6 | 73.4 | 81.7 | 78.6 | 77.2 | 74.2 | 75.6 | 74.8 | 73.4 | 72.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HEDNetFrame setting=single-frame, Model stages=one-stage2023.10 | 72.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 75.3 | 73.4 | 81.1 | 80.6 | 73.2 | 84.4 | 80 | 76.8 | 72.6 | 78.7 | 77.7 | 75.8 | 74.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HEDNetFPS=13.2, Params=12.64M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 72.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 81.1 | 80.6 | 73.2 | 84.4 | 80 | 76.8 | 72.6 | 78.7 | 77.7 | 75.8 | 74.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HEDNetOperator=SpConv, Input frames=12025.06 | 72.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 75.3 | 73.4 | 81.1 | 80.6 | 73.2 | 84.4 | 80 | 76.8 | 72.6 | 78.7 | 77.7 | 75.8 | 74.9 | 81.4 | 79.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SAFDNetFPS=13.6, Params=9.89M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 72.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.6 | 80.1 | 72.7 | 84.7 | 80.4 | 77.3 | 73.1 | 80 | 79 | 77.2 | 76.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Voxel MambaFPS=8.4, Params=20.99M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 72.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.8 | 80.3 | 72.6 | 85 | 80.8 | 77.7 | 73.6 | 78.6 | 77.6 | 75.7 | 74.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SAFDNet (reproduced)FPS=13.6, Params=9.89M, Reproduced result (marked with †)=true, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 72.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.6 | 80.2 | 72.7 | 85 | 80.7 | 77.7 | 73.4 | 79.3 | 78.2 | 76.4 | 75.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PointSlice-W-conv5FPS=14.6, Params=8.43M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 72.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.4 | 80 | 72.5 | 84.3 | 79.7 | 76.8 | 72.4 | 78.3 | 77.3 | 75.5 | 74.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PointSlice-WFPS=15.4, Params=7.82M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 71.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.3 | 79.8 | 72.3 | 84.1 | 79.4 | 76.4 | 71.9 | 77.9 | 76.9 | 75.1 | 74.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNN++number of point cloud frames=22021.01 | 71.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.17 | 79.7 | 72.14 | 83.48 | 80.42 | 75.54 | 72.61 | 74.63 | 73.75 | 72.35 | 71.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LION-Mamba-LOperator=RNN, Stages=One-stage, doubled channel dimension=true2024.07 | 71.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 75.9 | 74 | 80.3 | 79.9 | 72 | 85.8 | 81.4 | 78.5 | 74.3 | 80.1 | 79 | 77.2 | 76.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LIONFPS=4.9, Params=16.07M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 71.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.3 | 79.9 | 72 | 85.8 | 81.4 | 78.5 | 74.3 | 80.1 | 79 | 77.2 | 76.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-34Stages=One, Input Frames=Two-frame2022.05 | 71.53 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.98 | 79.47 | 72 | 82.52 | 79.33 | 75 | 71.95 | 70.51 | 69.69 | 68.38 | 67.58 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Voxel-or-PillarFPS=17.2, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 71.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.2 | 79.7 | 71.9 | 82.5 | 76.9 | 74.8 | 69.4 | 77.1 | 76 | 74.2 | 73.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CenterPoint++#Frames=3, #MACs (G)=113, Latency (ms)=13.6, Speedup (w.r.t. [94])=1.3x2023.01 | 71.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.6 | — | — | 71.8 | — | — | 73.5 | 70.8 | — | — | 73.7 | 72.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LitePTInput frames=single, Difficulty level=L22025.12 | 71.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 70.7 | — | — | 71.6 | — | — | 76.1 | 70.1 | — | — | 71.8 | 70.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet#Frames=2, #MACs (G)=148.8, Latency (ms)=11.6, Speedup (w.r.t. [94])=1.4x2023.01 | 71.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 70 | — | — | 71.6 | — | — | 74.5 | 71.4 | — | — | 68.3 | 67.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FastPillarsOperator=Conv, Input frames=12025.06 | 71.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 69.3 | — | — | 71.5 | — | — | 73.2 | 67.2 | — | — | 70.5 | 69.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-18Stages=One, Input Frames=Two-frame2022.05 | 71.08 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.59 | 79.06 | 71.56 | 82.11 | 78.82 | 74.49 | 71.35 | 70.41 | 69.57 | 68.27 | 67.46 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FlatFormer#Frames=3, #MACs (G)=193.2, Latency (ms)=12.7, Speedup (w.r.t. [94])=1.4x2023.01 | 71 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 72 | — | — | 71.4 | — | — | 74.5 | 71.3 | — | — | 74.7 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSVT-VoxelFrame setting=single-frame, Model stages=one-stage2023.10 | 71 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 74 | 72.1 | 79.7 | 79.3 | 71.4 | 83.7 | 78.9 | 76.1 | 71.5 | 77.5 | 76.5 | 74.6 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LION-RetNetOperator=RNN, Stages=One-stage2024.07 | 71 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 74 | 72.1 | 79.7 | 79.3 | 71.4 | 83.7 | 78.9 | 76.1 | 71.5 | 77.5 | 76.5 | 74.6 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LION-MambaOperator=RNN, Stages=One-stage2024.07 | 71 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 74.7 | 72.8 | 79.7 | 79.3 | 71.3 | 84.6 | 80 | 77.1 | 72.7 | 78.7 | 77.7 | 75.8 | 74.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSVT-VoxelFPS=8.4, Params=9.10M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 71 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.7 | 79.3 | 71.4 | 83.7 | 78.9 | 76.1 | 71.5 | 77.5 | 76.5 | 74.6 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FSDv2FPS=9.5, Params=44.83M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 71 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.8 | 79.3 | 71.4 | 84.8 | 79.7 | 77.4 | 72.5 | 80.7 | 79.6 | 77.9 | 76.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PTv3Input frames=single, Difficulty level=L22025.12 | 70.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 70.5 | — | — | 71.2 | — | — | 76.3 | 70.4 | — | — | 71.5 | 70.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PTv3FPS=12.5, Params=5.23M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 70.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.2 | — | — | 76.3 | 70.4 | — | — | 71.5 | 70.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PTv3Operator=SpConv, Input frames=12025.06 | 70.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 70.5 | — | — | 71.2 | — | — | 76.3 | 70.4 | — | — | 71.5 | 70.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SWFormer-3fFrames=32023.03 | 70.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.4 | 78.9 | 71.1 | 82.9 | 79 | 74.8 | 71.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SWFormer#Frames=3, Latency (ms)=20, Speedup (w.r.t. [94])=0.9x2023.01 | 70.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.1 | — | — | 74.8 | 71.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-34Input type=single-frame point cloud, Test-time augmentations=false, Model ensemble=false2022.07 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71 | 68.5 | 79.1 | 78.6 | 70.9 | 80.6 | 74 | 72.3 | 66.2 | 72.3 | 71.2 | 69.7 | 68.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-34Frame setting=single-frame, Model stages=one-stage2023.10 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71 | 68.5 | 79.1 | 78.6 | 70.9 | 80.6 | 74 | 72.3 | 66.2 | 72.3 | 71.2 | 69.7 | 68.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-34Operator=Sparse Conv, Stages=One-stage2024.07 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71 | 68.5 | 79.1 | 78.6 | 70.9 | 80.6 | 74 | 72.3 | 66.2 | 72.3 | 71.2 | 69.7 | 68.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSVT-VoxelOperator=Transformer, Stages=One-stage2024.07 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 73.2 | 71 | 79.3 | 78.8 | 70.9 | 82.8 | 77 | 75.2 | 69.8 | 76.4 | 75.4 | 73.6 | 72.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-34Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.1 | 78.6 | 70.9 | 80.6 | 74 | 72.3 | 66.2 | 72.3 | 71.2 | 69.7 | 68.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-34Operator=SpConv, Input frames=12025.06 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71 | 68.5 | 79.1 | 78.6 | 70.9 | 80.6 | 74 | 72.3 | 66.2 | 72.3 | 71.2 | 69.7 | 68.7 | 77.3 | 74.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSVT-PillarOperator=Transformer, Input frames=12025.06 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 73.2 | 71 | 79.3 | 78.8 | 70.9 | 82.8 | 77 | 75.2 | 69.8 | 76.4 | 75.4 | 73.6 | 72.7 | 79.5 | 77.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNetFrames=12023.03 | 70.46 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 74.6 | — | — | — | — | — | — | — | — | 68.43 | 79.09 | 78.59 | 70.92 | 80.59 | 74.01 | 72.28 | 66.17 | 72.29 | 71.21 | 69.72 | 68.67 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-34Stages=One, Input Frames=Single-frame2022.05 | 70.46 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.09 | 78.59 | 70.92 | 80.59 | 74.01 | 72.28 | 66.17 | 72.29 | 71.21 | 69.72 | 68.67 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FlatFormer#Frames=2, #MACs (G)=186.6, Latency (ms)=11.9, Speedup (w.r.t. [94])=1.4x2023.01 | 70.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.2 | — | — | 70.8 | — | — | 73.8 | 70.5 | — | — | 73.6 | 72.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNN++(center)Input type=single-frame point cloud, Test-time augmentations=false, Model ensemble=false2022.07 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.7 | 69.5 | 79.3 | 78.8 | 70.6 | 81.3 | 76.3 | 73.2 | 68 | 73.7 | 72.7 | 71.2 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNN++Frame setting=single-frame, Model stages=two-stage2023.10 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.7 | 69.5 | 79.3 | 78.8 | 70.6 | 81.3 | 76.3 | 73.2 | 68 | 73.7 | 72.7 | 71.2 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNN+++Operator=Sparse Conv, Stages=Two-stage2024.07 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.7 | 69.5 | 79.3 | 78.8 | 70.6 | 81.3 | 76.3 | 73.2 | 68 | 73.7 | 72.7 | 71.2 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LION-RWKVOperator=RNN, Stages=One-stage2024.07 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 74.6 | 72.7 | 79 | 78.5 | 70.6 | 84.6 | 80 | 77.2 | 72.8 | 79 | 78 | 76.1 | 75.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNN++FPS=7.4, Params=16.10M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.3 | 78.8 | 70.6 | 81.3 | 76.3 | 73.2 | 68 | 73.7 | 72.7 | 71.2 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNN++number of point cloud frames=12021.01 | 70.18 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.25 | 78.78 | 70.61 | 81.83 | 76.28 | 73.17 | 68 | 73.72 | 72.66 | 71.21 | 70.19 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FSDspconvInput type=single-frame point cloud, Test-time augmentations=false, Model ensemble=false, Schedule=12 epochs2022.07 | 70.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 72.9 | 70.8 | 79.2 | 78.8 | 70.5 | 82.6 | 77.3 | 73.9 | 69.1 | 77.1 | 76 | 74.4 | 73.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FSDOperator=Sparse Conv, Stages=Two-stage2024.07 | 70.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 72.9 | 70.8 | 79.2 | 78.8 | 70.5 | 82.6 | 77.3 | 73.9 | 69.1 | 77.1 | 76 | 74.4 | 73.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FSDv1FPS=7.1, Params=24.31M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 70.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.2 | 78.8 | 70.5 | 82.6 | 77.3 | 73.9 | 69.1 | 77.1 | 76 | 74.4 | 73.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-vggStages=One, Input Frames=Two-frame2022.05 | 70.07 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78.26 | 77.73 | 70.56 | 80.88 | 77.53 | 72.73 | 69.58 | 67.72 | 66.88 | 65.54 | 64.72 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-18Stages=One, Input Frames=Single-frame2022.05 | 69.92 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78.24 | 77.73 | 70.4 | 79.8 | 72.59 | 71.57 | 64.9 | 70.4 | 69.29 | 67.75 | 66.68 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNN++Frames=12023.03 | 69.91 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 75.21 | — | — | — | — | — | — | — | — | 68.61 | 79.1 | 78.63 | 70.34 | 80.62 | 74.62 | 71.86 | 66.3 | 73.49 | 72.38 | 70.7 | 69.62 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet#Frames=1, #MACs (G)=138.3, Latency (ms)=11.1, Speedup (w.r.t. [94])=1.3x2023.01 | 69.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 67.2 | — | — | 70.4 | — | — | 71.6 | 64.9 | — | — | 67.8 | 66.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FSDFrame setting=single-frame, Model stages=two-stage2023.10 | 69.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 72.7 | 70.5 | 79.5 | 79 | 70.3 | 83.6 | 78.2 | 74.4 | 69.4 | 75.3 | 74.1 | 73.3 | 72.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNetInput frames=single, Difficulty level=L22025.12 | 69.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 67.2 | — | — | 70.4 | — | — | 71.6 | 64.9 | — | — | 67.8 | 66.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-18FPS=22.7, Params=13.42M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 69.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78.2 | 77.7 | 70.4 | 79.8 | 72.6 | 71.6 | 64.9 | 70.4 | 69.3 | 67.8 | 66.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNeXtOperator=Sparse Conv, Stages=One-stage2024.07 | 69.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.9 | 69.7 | 78.4 | 77.9 | 70.3 | 82.5 | 77.1 | 74.9 | 69.8 | 73.2 | 72.2 | 70.6 | 69.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNeXtFPS=7.7, Params=9.47M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 69.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78.4 | 77.9 | 70.3 | 82.5 | 77.1 | 74.9 | 69.8 | 73.2 | 72.2 | 70.6 | 69.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNeXtOperator=SpConv, Input frames=12025.06 | 69.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.9 | 69.7 | 78.4 | 77.9 | 70.3 | 82.5 | 77.1 | 74.9 | 69.8 | 73.2 | 72.2 | 70.6 | 69.6 | 78 | 75.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNN++Input type=single-frame point cloud, Test-time augmentations=false, Model ensemble=false2022.07 | 69.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 68.4 | 64.9 | 78.8 | 78.2 | 70.3 | 76.7 | 67.2 | 68.5 | 59.7 | 69 | 67.6 | 66.5 | 65.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FALOOperator=Conv, MLP, HP, Input frames=12025.06 | 69.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 73.6 | 70.6 | 78.7 | 78.3 | 70.3 | 82.6 | 76.9 | 75 | 69.6 | 75.4 | 74.4 | 75.4 | 72.5 | 78.9 | 76.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SST-3fFrames=32023.03 | 69.57 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78.66 | 78.21 | 69.98 | 83.81 | 80.14 | 75.94 | 72.37 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CenterFormerFrame setting=single-frame, Model stages=one-stage2023.10 | 69.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.1 | 68.9 | 75 | 74.4 | 69.9 | 78.6 | 73 | 73.6 | 68.3 | 72.3 | 71.3 | 69.8 | 68.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LargeKernel3DFrame setting=single-frame, Model stages=one-stage2023.10 | 69.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78.1 | 77.6 | 69.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoxelNextOperator=Sparse Conv, Stages=One-stage2024.07 | 69.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 72.2 | 70.1 | 78.2 | 77.7 | 69.9 | 81.5 | 76.3 | 73.5 | 68.6 | 76.1 | 74.9 | 73.3 | 72.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CenterFormerOperator=Sparse Conv, Stages=One-stage2024.07 | 69.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.1 | 68.9 | 75 | 74.4 | 69.9 | 78.6 | 73 | 73.6 | 68.3 | 72.3 | 71.3 | 69.8 | 68.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CenterFormerFrame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 69.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 75 | 74.4 | 69.9 | 78 | 72.4 | 73.1 | 67.7 | 73.8 | 72.7 | 71.3 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoxelNeXtFPS=7.9, Params=34.01M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 69.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78.2 | 77.7 | 69.9 | 81.5 | 76.3 | 73.5 | 68.6 | 76.1 | 74.9 | 73.3 | 72.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoxelNextOperator=SpConv, Input frames=12025.06 | 69.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 72.2 | 70.1 | 78.2 | 77.7 | 69.9 | 81.5 | 76.3 | 73.5 | 68.6 | 76.1 | 74.9 | 73.3 | 72.2 | 78.6 | 76.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| OcTrFrame setting=single-frame, Model stages=one-stage2023.10 | 69.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 70.7 | 68.2 | 78.1 | 77.6 | 69.8 | 80.8 | 74.4 | 72.5 | 66.5 | 72.6 | 71.5 | 69.9 | 68.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSVT-PillarOperator=Transformer, Stages=One-stage2024.07 | 69.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 70.7 | 68.2 | 78.1 | 77.6 | 69.8 | 80.8 | 74.4 | 72.5 | 66.5 | 72.6 | 71.5 | 69.9 | 68.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| OcTrOperator=Transformer, Input frames=12025.06 | 69.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 70.7 | 68.2 | 78.1 | 77.6 | 69.8 | 80.8 | 74.4 | 72.5 | 66.5 | 72.6 | 71.5 | 69.9 | 68.9 | 77.2 | 74.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AFDetV2Stages=One, Input Frames=Single-frame2022.05 | 69.22 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.64 | 77.14 | 69.68 | 80.19 | 74.62 | 72.16 | 66.95 | 73.72 | 72.74 | 71.06 | 70.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AFDetV2Input type=single-frame point cloud, Test-time augmentations=false, Model ensemble=false2022.07 | 69.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71 | 68.8 | 77.6 | 77.1 | 69.7 | 80.2 | 74.6 | 72.2 | 67 | 73.7 | 72.7 | 71 | 70.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AFDetV2Frame setting=single-frame, Model stages=one-stage2023.10 | 69.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71 | 68.8 | 77.6 | 77.1 | 69.7 | 80.2 | 74.6 | 72.2 | 67 | 73.7 | 72.7 | 71 | 70.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AFDetV2Operator=Sparse Conv, Stages=One-stage2024.07 | 69.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71 | 68.5 | 77.6 | 77.1 | 69.7 | 80.2 | 74.6 | 72.2 | 67 | 73.7 | 72.7 | 71 | 70.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AFDetV2FPS=22.2, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 69.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.6 | 77.1 | 69.7 | 80.2 | 74.6 | 72.2 | 67 | 73.7 | 72.7 | 71 | 70.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RSNnumber of point cloud frames=32021.01 | 69.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78.4 | 78.1 | 69.5 | 79.4 | 76.2 | 69.9 | 67 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UPIDet2023.01 | 69.04 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78.36 | 77.91 | 69.45 | 76.32 | 71.67 | 65.93 | 61.81 | 79.64 | 78.55 | 76.36 | 75.26 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNNFrame setting=single-frame, Model stages=two-stage2023.10 | 69 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 69.6 | 67.2 | 78 | 77.5 | 69.4 | 79.2 | 73 | 70.4 | 64.7 | 71.5 | 70.3 | 69 | 67.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNNOperator=Sparse Conv, Stages=Two-stage2024.07 | 69 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 69.6 | 67.2 | 78 | 77.5 | 69.4 | 79.2 | 73 | 70.4 | 64.7 | 71.5 | 70.3 | 69 | 67.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNNFPS=3.1, Params=13.50M, Frame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 69 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78 | 77.5 | 69.4 | 79.2 | 73 | 70.4 | 64.7 | 71.5 | 70.3 | 69 | 67.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNNnumber of point cloud frames=12021.01 | 68.98 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78 | 77.5 | 69.43 | 79.21 | 73.03 | 70.42 | 64.72 | 71.46 | 70.27 | 68.95 | 67.79 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNet-vggStages=One, Input Frames=Single-frame2022.05 | 68.96 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.41 | 76.86 | 69.46 | 78.3 | 70.32 | 70 | 62.62 | 69.48 | 68.35 | 66.87 | 65.78 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SWFormer-1fFrames=12023.03 | 68.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.8 | 77.3 | 69.2 | 80.9 | 72.7 | 72.5 | 64.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SWFormerFrame setting=single-frame, Model stages=one-stage2023.10 | 68.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.8 | 77.3 | 69.2 | 80.9 | 72.7 | 72.5 | 64.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| OcTrOperator=Transformer, Stages=One-stage2024.07 | 68.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.1 | 68.9 | 77.8 | 77.3 | 69.2 | 80.9 | 72.7 | 72.5 | 64.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SWFormerFrame setting=single-frame, Hardware=NVIDIA 3090 GPU, Batch size=12025.09 | 68.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.8 | 77.3 | 69.2 | 80.9 | 72.7 | 72.5 | 64.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SWFormerOperator=Transformer, Input frames=12025.06 | 68.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.8 | 77.3 | 69.2 | 80.9 | 72.7 | 72.5 | 64.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FlatFormer#Frames=1, #MACs (G)=177.2, Latency (ms)=10.8, Speedup (w.r.t. [94])=1.4x2023.01 | 68.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 67.2 | — | — | 69 | — | — | 71.5 | 65.3 | — | — | 68.6 | 67.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FlatFormerInput frames=single, Difficulty level=L22025.12 | 68.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 67.2 | — | — | 69 | — | — | 71.5 | 65.3 | — | — | 68.6 | 67.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FlatFormerOperator=Transformer, Input frames=12025.06 | 68.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 67.2 | — | — | 69 | — | — | 71.5 | 65.3 | — | — | 68.6 | 67.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FSDspconvInput type=single-frame point cloud, Test-time augmentations=false, Model ensemble=false2022.07 | 68.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.9 | 69.7 | 77.8 | 77.3 | 68.9 | 81.9 | 76.4 | 73.2 | 68 | 76.5 | 75.2 | 73.8 | 72.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNNFrames=12023.03 | 68.41 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 69.63 | — | — | — | — | — | — | — | — | 63.33 | 77.51 | 76.89 | 68.98 | 75.01 | 65.65 | 66.04 | 57.61 | 67.81 | 66.35 | 65.39 | 63.98 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PV-RCNNStages=Two, Input Frames=Single-frame2022.05 | 68.41 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.51 | 76.89 | 68.98 | 75.01 | 65.65 | 66.04 | 57.61 | 67.81 | 66.35 | 65.39 | 63.98 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — |