def _assign_label_to_anchor(self, valid_anchor, gt_bbox, neg_thr, pos_thr, min_pos_thr):
num_anchor = valid_anchor.shape[0]
cls_label = np.full(shape=(num_anchor,), fill_value=-1, dtype=np.float32)
if len(gt_bbox) > 0:
# num_anchor x num_gt
overlaps = bbox_overlaps_cython(valid_anchor.astype(np.float32, copy=False), gt_bbox.astype(np.float32, copy=False))
max_overlaps = overlaps.max(axis=1)
argmax_overlaps = overlaps.argmax(axis=1)
gt_max_overlaps = overlaps.max(axis=0)
# TODO: speed up this
# TODO: fix potentially assigning wrong anchors as positive
# A correct implementation is given as
# gt_argmax_overlaps = np.where((overlaps.transpose() == gt_max_overlaps[:, None]) &
# (overlaps.transpose() >= min_pos_thr))[1]
gt_argmax_overlaps = np.where((overlaps == gt_max_overlaps) &
(overlaps >= min_pos_thr))[0]
# anchor class
cls_label[max_overlaps < neg_thr] = 0
# fg label: for each gt, anchor with highest overlap
cls_label[gt_argmax_overlaps] = 1
# fg label: above threshold IoU
cls_label[max_overlaps >= pos_thr] = 1
else:
cls_label[:] = 0
argmax_overlaps = np.zeros(shape=(num_anchor, ))
return cls_label, argmax_overlaps
def _assign_label_to_anchor_group(self, valid_anchor, gt_bbox, gt_class,
neg_thr, pos_thr, min_pos_thr):
num_anchor = valid_anchor.shape[0]
cls_label = np.full(shape=(num_anchor, ),
fill_value=-1,
dtype=np.float32)
if len(gt_bbox) > 0:
# num_anchor x num_gt
overlaps = bbox_overlaps_cython(
valid_anchor.astype(np.float32, copy=False),
gt_bbox.astype(np.float32, copy=False))
max_overlaps = overlaps.max(axis=1)
argmax_overlaps = overlaps.argmax(axis=1)
gt_max_overlaps = overlaps.max(axis=0)
mask_label_map = np.tile(
gt_class.reshape(-1).astype(np.float32),
num_anchor).reshape(num_anchor, -1)
max_overlaps_label = mask_label_map[np.arange(num_anchor),
argmax_overlaps]
gt_argmax_overlaps = np.where((overlaps == gt_max_overlaps)
& (overlaps >= min_pos_thr))
# anchor class
cls_label[max_overlaps < neg_thr] = 0
# fg label: for each gt, anchor with highest overlap
cls_label[
gt_argmax_overlaps[0]] = mask_label_map[gt_argmax_overlaps]
# fg label: above threshold IoU
fg_label_idxs = np.where(max_overlaps >= pos_thr)[0]
cls_label[fg_label_idxs] = max_overlaps_label[fg_label_idxs]
else:
cls_label[:] = 0
argmax_overlaps = np.zeros(shape=(num_anchor, ))
return cls_label, argmax_overlaps
def _filter_anchor_by_scale_range(self, cls_label, valid_anchor, gt_bbox,
valid_range, invalid_anchor_threshd):
if len(gt_bbox) == 0:
return
gt_bbox_sizes = (gt_bbox[:, 2] - gt_bbox[:, 0] +
1.0) * (gt_bbox[:, 3] - gt_bbox[:, 1] + 1.0)
invalid_gt_bbox_inds = np.where((gt_bbox_sizes < valid_range[0]**2) | (
gt_bbox_sizes > valid_range[1]**2))[0]
invalid_gt_bbox = gt_bbox[invalid_gt_bbox_inds]
if len(invalid_gt_bbox) > 0:
invalid_overlaps = bbox_overlaps_cython(
valid_anchor.astype(np.float32, copy=False),
invalid_gt_bbox.astype(np.float32, copy=False))
invalid_argmax_overlaps = invalid_overlaps.argmax(axis=1)
invalid_max_overlaps = invalid_overlaps[
np.arange(len(valid_anchor)), invalid_argmax_overlaps]
# ignore anchors overlapped with invalid gt boxes
disable_inds = np.where(
(invalid_max_overlaps > invalid_anchor_threshd))[0]
cls_label[disable_inds] = -1