Scale-Aware Attention-Based PillarsNet (SAPN) Based 3D Object Detection for Point Cloud

Table 1

3D detection performance: average precision (AP) for car objects in easy, moderate, and hard splits of the KITTI validation set.

Method	Modality	Easy	Mod	Hard
MV3D [4]	RGB + LiDAR	71.29	62.68	56.56
F-PointNet [16]	RGB + LiDAR	83.76	70.92	63.65
AVOD-FPN [15]	RGB + LiDAR	84.41	74.44	68.65
VoxelNet [5]	LiDAR	81.97	65.46	62.85
SECOND [14]	LiDAR	87.43	76.48	69.10
PointPillars [6]	LiDAR	85.43	75.90	69.59
Ours	LiDAR	88.38	78.28	69.52