-
Notifications
You must be signed in to change notification settings - Fork 111
/
retinanet_loss.py
166 lines (141 loc) · 6.41 KB
/
retinanet_loss.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
"""
This file contains specific functions for computing losses on the RetinaNet
file
"""
import torch
from ..utils import cat
from maskrcnn_benchmark.layers import SmoothL1Loss
from maskrcnn_benchmark.layers import SigmoidFocalLoss
from maskrcnn_benchmark.modeling.matcher import Matcher
from maskrcnn_benchmark.structures.boxlist_ops import boxlist_iou
from maskrcnn_benchmark.structures.boxlist_ops import cat_boxlist
class RetinaNetLossComputation(object):
"""
This class computes the RetinaNet loss.
"""
def __init__(self, cfg, proposal_matcher, box_coder):
"""
Arguments:
proposal_matcher (Matcher)
box_coder (BoxCoder)
"""
# self.target_preparator = target_preparator
self.proposal_matcher = proposal_matcher
self.box_coder = box_coder
self.num_classes = cfg.RETINANET.NUM_CLASSES -1
self.box_cls_loss_func = SigmoidFocalLoss(
self.num_classes,
cfg.RETINANET.LOSS_GAMMA,
cfg.RETINANET.LOSS_ALPHA
)
self.regression_loss = SmoothL1Loss(
beta=cfg.RETINANET.BBOX_REG_BETA
)
def match_targets_to_anchors(self, anchor, target):
match_quality_matrix = boxlist_iou(target, anchor)
matched_idxs = self.proposal_matcher(match_quality_matrix)
# RPN doesn't need any fields from target
# for creating the labels, so clear them all
target = target.copy_with_fields(['labels'])
# get the targets corresponding GT for each anchor
# NB: need to clamp the indices because we can have a single
# GT in the image, and matched_idxs can be -2, which goes
# out of bounds
matched_targets = target[matched_idxs.clamp(min=0)]
matched_targets.add_field("matched_idxs", matched_idxs)
return matched_targets
def prepare_targets(self, anchors, targets):
labels = []
regression_targets = []
for anchors_per_image, targets_per_image in zip(anchors, targets):
matched_targets = self.match_targets_to_anchors(
anchors_per_image, targets_per_image
)
matched_idxs = matched_targets.get_field("matched_idxs")
labels_per_image = matched_targets.get_field("labels").clone()
# Background (negative examples)
bg_indices = matched_idxs == Matcher.BELOW_LOW_THRESHOLD
labels_per_image[bg_indices] = 0
# discard indices that are between thresholds
# -1 will be ignored in SigmoidFocalLoss
inds_to_discard = matched_idxs == Matcher.BETWEEN_THRESHOLDS
labels_per_image[inds_to_discard] = -1
labels_per_image = labels_per_image.to(dtype=torch.float32)
# compute regression targets
regression_targets_per_image = self.box_coder.encode(
matched_targets.bbox, anchors_per_image.bbox
)
labels.append(labels_per_image)
regression_targets.append(regression_targets_per_image)
return labels, regression_targets
def __call__(self, anchors, box_cls, box_regression, targets):
"""
Arguments:
anchors (list[BoxList])
objectness (list[Tensor])
box_regression (list[Tensor])
targets (list[BoxList])
Returns:
objectness_loss (Tensor)
box_loss (Tensor
"""
anchors = [cat_boxlist(anchors_per_image) for anchors_per_image in anchors]
labels, regression_targets = self.prepare_targets(anchors, targets)
# sampled_pos_inds, sampled_neg_inds = self.fg_bg_sampler(labels)
# sampled_pos_inds = torch.nonzero(torch.cat(sampled_pos_inds, dim=0)).squeeze(1)
# sampled_neg_inds = torch.nonzero(torch.cat(sampled_neg_inds, dim=0)).squeeze(1)
# sampled_inds = torch.cat([sampled_pos_inds, sampled_neg_inds], dim=0)
num_layers = len(box_cls)
box_cls_flattened = []
box_regression_flattened = []
# for each feature level, permute the outputs to make them be in the
# same format as the labels. Note that the labels are computed for
# all feature levels concatenated, so we keep the same representation
# for the objectness and the box_regression
for box_cls_per_level, box_regression_per_level in zip(
box_cls, box_regression
):
N, A, H, W = box_cls_per_level.shape
C = self.num_classes
box_cls_per_level = box_cls_per_level.view(N, -1, C, H, W)
box_cls_per_level = box_cls_per_level.permute(0, 3, 4, 1, 2)
box_cls_per_level = box_cls_per_level.reshape(N, -1, C)
box_regression_per_level = box_regression_per_level.view(N, -1, 4, H, W)
box_regression_per_level = box_regression_per_level.permute(0, 3, 4, 1, 2)
box_regression_per_level = box_regression_per_level.reshape(N, -1, 4)
box_cls_flattened.append(box_cls_per_level)
box_regression_flattened.append(box_regression_per_level)
# concatenate on the first dimension (representing the feature levels), to
# take into account the way the labels were generated (with all feature maps
# being concatenated as well)
box_cls = cat(box_cls_flattened, dim=1).reshape(-1, C)
box_regression = cat(box_regression_flattened, dim=1).reshape(-1, 4)
labels = torch.cat(labels, dim=0)
regression_targets = torch.cat(regression_targets, dim=0)
pos_inds = labels > 0
retinanet_regression_loss = self.regression_loss(
box_regression[pos_inds],
regression_targets[pos_inds],
size_average=False,
) / (pos_inds.sum() * 4)
labels = labels.int()
retinanet_cls_loss =self.box_cls_loss_func(
box_cls,
labels
) / ((labels > 0).sum() + N)
losses = {
"loss_retina_cls": retinanet_cls_loss,
"loss_retina_reg": retinanet_regression_loss,
}
return losses
def make_retinanet_loss_evaluator(cfg, box_coder):
matcher = Matcher(
cfg.MODEL.RPN.FG_IOU_THRESHOLD,
cfg.MODEL.RPN.BG_IOU_THRESHOLD,
allow_low_quality_matches=cfg.RETINANET.LOW_QUALITY_MATCHES,
low_quality_threshold=cfg.RETINANET.LOW_QUALITY_THRESHOLD
)
loss_evaluator = RetinaNetLossComputation(
cfg, matcher, box_coder
)
return loss_evaluator