def _compute_targets(ex_rois, gt_rois):
assert ex_rois.shape[0] == gt_rois.shape[0]
assert ex_rois.shape[1] == 4
assert gt_rois.shape[1] == 5
return bbox_transform(ex_rois, gt_rois[:, :4]).astype(np.float32,
copy=False)
def _compute_targets(ex_rois, gt_rois):
"""Compute bounding-box regression targets for an image."""
assert ex_rois.shape[0] == gt_rois.shape[0]
assert ex_rois.shape[1] == 4
assert gt_rois.shape[1] == 5
return bbox_transform(ex_rois, gt_rois[:, :4]).astype(np.float32, copy=False)
def _compute_targets(ex_rois, gt_rois):
"""Compute bounding-box regression targets for an image.
gt_rois（shape: [:, 8]）: 是label box:[:4] indx:[4] dis:[5:]
"""
assert ex_rois.shape[0] == gt_rois.shape[0]
assert ex_rois.shape[1] == 4
assert gt_rois.shape[1] == 8
return bbox_transform(ex_rois, gt_rois[:, :4]).astype(np.float32,
copy=False)
def _compute_targets(anchor_rois, gt_rois):
"""
Compute bounding box regression targets for an image
:param anchor_rois <torch.Tensor>:
:param gt_rois:
:return:
"""
assert anchor_rois.shape[0] == gt_rois.shape[0]
assert anchor_rois.shape[1] == 6
assert gt_rois.shape[1] == 7
return bbox_transform(anchor_rois, gt_rois[:, :6])
def _compute_targets(anchor_rois, gt_rois):
"""
Compute bounding box regression targets for an image
:param anchor_rois <torch.Tensor>:
:param gt_rois:
:return:
"""
assert anchor_rois.shape[0] == gt_rois.shape[0]
assert anchor_rois.shape[1] == 6
assert gt_rois.shape[1] == 7
return bbox_transform(anchor_rois, gt_rois[:,:6])
def _compute_targets(ex_rois, gt_rois, labels):
"""Compute bounding-box regression targets for an image."""
assert ex_rois.shape[0] == gt_rois.shape[0]
assert ex_rois.shape[1] == 4
assert gt_rois.shape[1] == 4
targets = bbox_transform(ex_rois, gt_rois)
if cfg.FLAGS.bbox_normalize_targets_precomputed:
# Optionally normalize targets by a precomputed mean and stdev
targets = ((targets - np.array(cfg.FLAGS2["bbox_normalize_means"]))
/ np.array(cfg.FLAGS2["bbox_normalize_stds"]))
return np.hstack(
(labels[:, np.newaxis], targets)).astype(np.float32, copy=False)
def _compute_targets(ex_rois, gt_rois, labels):
"""
Compute bounding box regression targets for an image
Inputs are tensor
:param ex_rois:
:param gt_rois:
:param labels:
:return:
"""
assert ex_rois.shape[0] == gt_rois.shape[0]
assert ex_rois.shape[1] == 6
assert gt_rois.shape[1] == 6 or gt_rois.shape[1] == 12
targets = bbox_transform(ex_rois, gt_rois)
return torch.cat([targets, labels.unsqueeze(1)], 1)
def _compute_targets(ex_rois, gt_rois, labels):
"""Compute bounding-box regression targets for an image."""
assert ex_rois.shape[0] == gt_rois.shape[0]
assert ex_rois.shape[1] == 4
assert gt_rois.shape[1] == 4
targets = bbox_transform(ex_rois, gt_rois)
# 对bbox偏移量进行归一化操作，扩大了10倍
if cfg.FLAGS.bbox_normalize_targets_precomputed: # True
targets = ((targets - np.array(cfg.FLAGS2["bbox_normalize_means"])
) # (0.0, 0.0, 0.0, 0.0)
/ np.array(cfg.FLAGS2["bbox_normalize_stds"])
) # (0.1, 0.1, 0.1, 0.1)
# 将lable拼接到bbox的第一列
return np.hstack((labels[:, np.newaxis], targets)).astype(np.float32,
copy=False)
def _compute_targets(ex_rois, gt_rois, labels):
"""
Compute bounding box regression targets for an image
Inputs are tensor
:param ex_rois:
:param gt_rois:
:param labels:
:return:
"""
assert ex_rois.shape[0] == gt_rois.shape[0]
assert ex_rois.shape[1] == 6
assert gt_rois.shape[1] == 6 or gt_rois.shape[1] == 12
targets = bbox_transform(ex_rois, gt_rois)
return torch.cat([targets, labels.unsqueeze(1)], 1)
