3D Object Detection on Lyft v1.0 (val)
79.6AP BEV (0-30m)Sup. (Lyft)
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Sup. (Lyft)IoU threshold=0.252022.03 | 79.6 | 77.5 | 66.4 | 64.4 | 47.8 | 43.8 | 65.5 | 63.2 | |
| Sup. (KITTI)IoU threshold=0.252022.03 | 72.3 | 69.5 | 53.2 | 48.1 | 27.9 | 20.5 | 53.1 | 48.1 | |
| MODEST (R40)Backbone=VoxelNet (SECOND), Number of Rounds (R)=40, IoU threshold=0.252022.03 | 61.1 | 56.2 | 57.5 | 53.4 | 41.2 | 29.8 | 54.1 | 47.6 | |
| MODEST (R10)Backbone=VoxelNet (SECOND), Number of Rounds (R)=10, IoU threshold=0.252022.03 | 56.8 | 51.3 | 51.4 | 40.5 | 19.2 | 9 | 44.1 | 35.5 | |
| MODEST (R0)Backbone=VoxelNet (SECOND), Number of Rounds (R)=0, IoU threshold=0.252022.03 | 44.9 | 40.4 | 24.5 | 14.8 | 2.7 | 0.7 | 26.3 | 19.8 | |
| MODEST-PP (R10)Backbone=PointPillars, Number of Rounds (R)=10, IoU threshold=0.252022.03 | 42.1 | 38.3 | 21.9 | 19.2 | 1 | 0.9 | 22.8 | 20.6 | |
| MODEST-PP (R0)Backbone=PointPillars, Number of Rounds (R)=0, IoU threshold=0.252022.03 | 34.1 | 31.3 | 5.1 | 3 | 0 | 0 | 12 | 9.7 |