L1 regression loss for proposals going to 0

[peiherng97]

Hi, I noticed that the smooth L1 regression loss for the fast r cnn goes to 0 when there are no proposals that have an iou of > 0.5. Is this deliberately done? The fg_indices are empty when there are no matches

easy-faster-rcnn.pytorch/model.py

Lines 124 to 129 in 2c30c6d

	if len(fg_masks.nonzero()) > 0:
	labels[fg_masks] = gt_classes_batch[fg_masks.nonzero()[:, 0], proposal_assignments[fg_masks]]
	# select 128 x `batch_size` samples
	fg_indices = (labels > 0).nonzero()
	bg_indices = (labels == 0).nonzero()