Skip to main content
Journal of Big Data

Table 4 Decrease in performance on the low-resolution point cloud dataset

From: Out-of-distribution- and location-aware PointNets for real-time 3D road user detection without a GPU

Method

Cars \(\Delta\)AP

Pedestrians \(\Delta\)AP

VoxelNet [3]

64.39

32.99

SECOND [4]

63.02

31.59

IA-SSD [15]

48.13

19.10

PointPillars [9]

59.96

28.85

Ours

23.28

26.35

  1. \(\Delta\)AP \(=\) (KITTI AP) − (16chn-LiDAR AP) illustrates the performance gap for models trained with the KITTI dataset and tested with the 16-channel LiDAR mobile robot dataset. Smaller is better
Back to article page