File size: 7,923 Bytes
51f6859
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
# Copyright (c) OpenMMLab. All rights reserved.
import torch
import torch.nn as nn
import torch.nn.functional as F

from ..builder import LOSSES
from .utils import weight_reduce_loss


def _expand_onehot_labels(labels, label_weights, label_channels):
    bin_labels = labels.new_full((labels.size(0), label_channels), 0)
    inds = torch.nonzero(
        (labels >= 0) & (labels < label_channels), as_tuple=False).squeeze()
    if inds.numel() > 0:
        bin_labels[inds, labels[inds]] = 1
    bin_label_weights = label_weights.view(-1, 1).expand(
        label_weights.size(0), label_channels)
    return bin_labels, bin_label_weights


# TODO: code refactoring to make it consistent with other losses
@LOSSES.register_module()
class GHMC(nn.Module):
    """GHM Classification Loss.

    Details of the theorem can be viewed in the paper
    `Gradient Harmonized Single-stage Detector
    <https://arxiv.org/abs/1811.05181>`_.

    Args:
        bins (int): Number of the unit regions for distribution calculation.
        momentum (float): The parameter for moving average.
        use_sigmoid (bool): Can only be true for BCE based loss now.
        loss_weight (float): The weight of the total GHM-C loss.
        reduction (str): Options are "none", "mean" and "sum".
            Defaults to "mean"
    """

    def __init__(self,
                 bins=10,
                 momentum=0,
                 use_sigmoid=True,
                 loss_weight=1.0,
                 reduction='mean'):
        super(GHMC, self).__init__()
        self.bins = bins
        self.momentum = momentum
        edges = torch.arange(bins + 1).float() / bins
        self.register_buffer('edges', edges)
        self.edges[-1] += 1e-6
        if momentum > 0:
            acc_sum = torch.zeros(bins)
            self.register_buffer('acc_sum', acc_sum)
        self.use_sigmoid = use_sigmoid
        if not self.use_sigmoid:
            raise NotImplementedError
        self.loss_weight = loss_weight
        self.reduction = reduction

    def forward(self,
                pred,
                target,
                label_weight,
                reduction_override=None,
                **kwargs):
        """Calculate the GHM-C loss.

        Args:
            pred (float tensor of size [batch_num, class_num]):
                The direct prediction of classification fc layer.
            target (float tensor of size [batch_num, class_num]):
                Binary class target for each sample.
            label_weight (float tensor of size [batch_num, class_num]):
                the value is 1 if the sample is valid and 0 if ignored.
            reduction_override (str, optional): The reduction method used to
                override the original reduction method of the loss.
                Defaults to None.
        Returns:
            The gradient harmonized loss.
        """
        assert reduction_override in (None, 'none', 'mean', 'sum')
        reduction = (
            reduction_override if reduction_override else self.reduction)
        # the target should be binary class label
        if pred.dim() != target.dim():
            target, label_weight = _expand_onehot_labels(
                target, label_weight, pred.size(-1))
        target, label_weight = target.float(), label_weight.float()
        edges = self.edges
        mmt = self.momentum
        weights = torch.zeros_like(pred)

        # gradient length
        g = torch.abs(pred.sigmoid().detach() - target)

        valid = label_weight > 0
        tot = max(valid.float().sum().item(), 1.0)
        n = 0  # n valid bins
        for i in range(self.bins):
            inds = (g >= edges[i]) & (g < edges[i + 1]) & valid
            num_in_bin = inds.sum().item()
            if num_in_bin > 0:
                if mmt > 0:
                    self.acc_sum[i] = mmt * self.acc_sum[i] \
                        + (1 - mmt) * num_in_bin
                    weights[inds] = tot / self.acc_sum[i]
                else:
                    weights[inds] = tot / num_in_bin
                n += 1
        if n > 0:
            weights = weights / n

        loss = F.binary_cross_entropy_with_logits(
            pred, target, reduction='none')
        loss = weight_reduce_loss(
            loss, weights, reduction=reduction, avg_factor=tot)
        return loss * self.loss_weight


# TODO: code refactoring to make it consistent with other losses
@LOSSES.register_module()
class GHMR(nn.Module):
    """GHM Regression Loss.

    Details of the theorem can be viewed in the paper
    `Gradient Harmonized Single-stage Detector
    <https://arxiv.org/abs/1811.05181>`_.

    Args:
        mu (float): The parameter for the Authentic Smooth L1 loss.
        bins (int): Number of the unit regions for distribution calculation.
        momentum (float): The parameter for moving average.
        loss_weight (float): The weight of the total GHM-R loss.
        reduction (str): Options are "none", "mean" and "sum".
            Defaults to "mean"
    """

    def __init__(self,
                 mu=0.02,
                 bins=10,
                 momentum=0,
                 loss_weight=1.0,
                 reduction='mean'):
        super(GHMR, self).__init__()
        self.mu = mu
        self.bins = bins
        edges = torch.arange(bins + 1).float() / bins
        self.register_buffer('edges', edges)
        self.edges[-1] = 1e3
        self.momentum = momentum
        if momentum > 0:
            acc_sum = torch.zeros(bins)
            self.register_buffer('acc_sum', acc_sum)
        self.loss_weight = loss_weight
        self.reduction = reduction

    # TODO: support reduction parameter
    def forward(self,
                pred,
                target,
                label_weight,
                avg_factor=None,
                reduction_override=None):
        """Calculate the GHM-R loss.

        Args:
            pred (float tensor of size [batch_num, 4 (* class_num)]):
                The prediction of box regression layer. Channel number can be 4
                or 4 * class_num depending on whether it is class-agnostic.
            target (float tensor of size [batch_num, 4 (* class_num)]):
                The target regression values with the same size of pred.
            label_weight (float tensor of size [batch_num, 4 (* class_num)]):
                The weight of each sample, 0 if ignored.
            reduction_override (str, optional): The reduction method used to
                override the original reduction method of the loss.
                Defaults to None.
        Returns:
            The gradient harmonized loss.
        """
        assert reduction_override in (None, 'none', 'mean', 'sum')
        reduction = (
            reduction_override if reduction_override else self.reduction)
        mu = self.mu
        edges = self.edges
        mmt = self.momentum

        # ASL1 loss
        diff = pred - target
        loss = torch.sqrt(diff * diff + mu * mu) - mu

        # gradient length
        g = torch.abs(diff / torch.sqrt(mu * mu + diff * diff)).detach()
        weights = torch.zeros_like(g)

        valid = label_weight > 0
        tot = max(label_weight.float().sum().item(), 1.0)
        n = 0  # n: valid bins
        for i in range(self.bins):
            inds = (g >= edges[i]) & (g < edges[i + 1]) & valid
            num_in_bin = inds.sum().item()
            if num_in_bin > 0:
                n += 1
                if mmt > 0:
                    self.acc_sum[i] = mmt * self.acc_sum[i] \
                        + (1 - mmt) * num_in_bin
                    weights[inds] = tot / self.acc_sum[i]
                else:
                    weights[inds] = tot / num_in_bin
        if n > 0:
            weights /= n
        loss = weight_reduce_loss(
            loss, weights, reduction=reduction, avg_factor=tot)
        return loss * self.loss_weight