3D Object Detection on KITTI Pedestrian (val)
61.15AP (Moderate)SOP2+LoRA
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SOP2+LoRATrainable Params=0.84M2025.12 | 61.15 | 66.34 | 56.2 | |
| SOP2Trainable Params=0.82M2025.12 | 61.12 | 67.24 | 56.43 | |
| Prompt GeneratorTrainable Params=1.64M2025.12 | 59.16 | 65.99 | 54.71 | |
| Full Fine-tuneTrainable Params=8.65M2025.12 | 55.81 | 60.44 | 52.04 | |
| Prompt TokenTrainable Params=0.45M2025.12 | 54.86 | 61.95 | 50.1 | |
| BitFitTrainable Params=0.47M2025.12 | 54.7 | 59.97 | 50.13 | |
| LoRATrainable Params=0.47M2025.12 | 51.87 | 56.66 | 49.11 | |
| Head Fine-tuneTrainable Params=0.45M2025.12 | 51.69 | 57.81 | 47.57 | |
| From ScratchTrainable Params=8.65M2025.12 | 48.54 | 53.47 | 45.2 | |
| DSGNModality=Stereo2020.01 | 33.85 | 40.16 | 29.43 | |
| PL: F-PointNet*Modality=Stereo, Pre-training=extra Scene Flow dataset2020.01 | 27.4 | 33.8 | 24 | |
| M3D-RPNModality=Mono2020.01 | 11.09 | — | — | |
| DEVIANT2022.07 | 7.18 | 9.85 | 5.42 | |
| MonoDIS-M2022.07 | 7.1 | 9.5 | 5.7 | |
| GUP Netmode=Retrained2022.07 | 6.84 | 9.37 | 5.73 | |
| GrooMeD-NMS2022.07 | 2.71 | 3.79 | 2.61 | |
| MonoDIS2022.07 | 2.28 | 3.2 | 1.71 |