3D Object Detection on WOD 100% training data (test)
81.7Vehicle L1 mAPOcTr (PV-RCNN++)
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OcTr (PV-RCNN++)Training Data=100%, Baseline=PV-RCNN++2023.03 | 81.7 | 81.4 | 74 | 73.6 | 81.2 | 75.2 | 75 | 69.3 | 73 | 71.8 | 70.4 | 69.4 | |
| PV-RCNN++Training Data=100%2023.03 | 81.6 | 81.2 | 73.9 | 73.5 | 80.4 | 75 | 74.1 | 69 | 71.9 | 70.8 | 69.3 | 68.2 | |
| OcTr (SECOND)Training Data=100%, Baseline=SECOND2023.03 | 77.9 | 77.4 | 70.2 | 69.7 | 71.5 | 61.1 | 65.7 | 56.1 | 70.7 | 69.3 | 68.1 | 66.8 | |
| SECONDTraining Data=100%2023.03 | 76.2 | 75.7 | 68.3 | 67.8 | 68.6 | 55.3 | 62.8 | 50.5 | 62.4 | 56.6 | 60.1 | 54.6 |