A Robust Convolutional Neural Network for 6D Object Pose Estimation from RGB Image with Distance Regularization Voting Loss

Table 5

The performance of our system in terms of the number of wins on the LINEMOD and occlusion datasets using ADD (-S) and 2D projection errors.

Number of wins	LINEMOD ADD (-S)	Occlusion ADD (-S)	LINEMOD 2D projection	Occlusion 2D projection
SSD6D [1]	1/13	0/8	0/13	0/8
PVNet [8]	0/13	0/8	1/13	2/8
DPVL [11]	2/13	0/8	2/13	0/8
PDAL + AFAM [12]	2/13	0/8	0/13	0/8
L+ [7]	2/13	3/8	1/13	1/8
Ours	11/13	5/8	11/13	5/8

The bold values given in this table show the highest numbers of wins on the LINEMOD and occlusion datasets using ADD (-S) and 2D projection errors.