System Performance Metrics on nuScenes (val)
15.2Training GPU Memory (GB)SparseDrive-S
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| SparseDrive-STraining Batch Size=6, Evaluation Hardware=NVIDIA Geforce RTX 4090 (8 GPUs for training, 1 GPU for inference)2024.05 | 15.2 | — | 1,294 | 192 | 85.9 | 9 | |
| SparseDrive-BTraining Batch Size=4, Evaluation Hardware=NVIDIA Geforce RTX 4090 (8 GPUs for training, 1 GPU for inference)2024.05 | 17.6 | — | 1,437 | 787 | 104.7 | 7.3 | |
| UniADTraining Batch Size=1, Evaluation Hardware=NVIDIA Tesla A100 (8 GPUs for training, 1 GPU for inference)2024.05 | 50 | — | 2,451 | 1,709 | 125 | 1.8 |