MANet: End-to-End Learning for Point Cloud Based on Robust Pointpillar and Multiattention

Table 4

Results on the test nuScenes 3D detection benchmark.

Methods	Car	Pedestrian	Cyclist	mAP
Baseline [6]	62.5	50.2	64.5	57.6
PVRCNN	64.8	46.7	—	—
Centerpoint	66.1	62.4	67.6	65.3
Point augmenting	62.2	64.6	73.3	66.7
Ours	65.1	63.7	65.9	64.9