Table 2.

Segmentation results of different regression models on the validation set. The highest IoUs are bolded

Mean IoU per image
$\alpha /\xi$	6	8	10	12
-2	0.593 ± 0.109	0.590 ± 0.102	0.576 ± 0.088	0.560 ± 0.066
-1	0.590 ± 0.105	0.579 ± 0.094	0.572 ± 0.089	0.566 ± 0.077
10-4	0.596 ± 0.115	0.584 ± 0.098	0.586 ± 0.097	0.584 ± 0.093
1	0.619 ± 0.125	0.601 ± 0.109	0.583 ± 0.105	0.587 ± 0.098
2	0.610 ± 0.123	0.588 ± 0.108	0.592 ± 0.105	0.580 ± 0.094

IoU per image
$\alpha /\xi$	6	8	10	12
-2	0.645 ± 0.207	0.645 ± 0.206	0.648 ± 0.200	0.626 ± 0.228
-1	0.647 ± 0.203	0.648 ± 0.201	0.643 ± 0.208	0.640 ± 0.212
10-4	0.648 ± 0.201	0.644 ± 0.207	0.641 ± 0.214	0.644 ± 0.207
1	0.649 ± 0.200	0.646 ± 0.206	0.647 ± 0.203	0.643 ± 0.208
2	0.650 ± 0.197	0.648 ± 0.202	0.649 ± 0.200	0.643 ± 0.208