retinaNet FocalLoss源码详解

• targets[positive_indices, assigned_annotations[positive_indices, 4].long()] = 1

## 把正样本所对应的锚框所对应的类别的列置为1
# aim = torch.randint(0, 1, (1, 80))
# tensor([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
#          0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
#          0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
#          0, 0, 0, 0, 0, 0, 0, 0]])
# aim[0, 12] = 1
# tensor([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
#          0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
#          0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
#          0, 0, 0, 0, 0, 0, 0, 0]])
1
2
3
4
5
6
7
8
9
10
11

• result = torch.where(condition, x, y)

import torch 
condition = torch.tensor([True, False, True, False]) 
x = torch.tensor([1, 2, 3, 4]) 
y = torch.tensor([10, 20, 30, 40])
result = torch.where(condition, x, y) 
print(result)
tensor([ 1, 20,  3, 40])
1
2
3
4
5
6
7

• torch.eq(targets, 1.)

targets = torch.tensor([1, 0, 1, 0, 1]) 
torch.eq(targets, 1.)
Out[20]: tensor([ True, False,  True, False,  True])
1
2
3

retinaNet FocalLoss源码详解

class FocalLoss(nn.Module):
    #def __init__(self):

    def forward(self, classifications, regressions, anchors, annotations):
        alpha = 0.25
        gamma = 2.0
        batch_size = classifications.shape[0]
        classification_losses = []
        regression_losses = []

        anchor = anchors[0, :, :]

        anchor_widths  = anchor[:, 2] - anchor[:, 0]
        anchor_heights = anchor[:, 3] - anchor[:, 1]
        anchor_ctr_x   = anchor[:, 0] + 0.5 * anchor_widths
        anchor_ctr_y   = anchor[:, 1] + 0.5 * anchor_heights

        for j in range(batch_size):

            classification = classifications[j, :, :]
            regression = regressions[j, :, :]

            bbox_annotation = annotations[j, :, :]
            bbox_annotation = bbox_annotation[bbox_annotation[:, 4] != -1]

            classification = torch.clamp(classification, 1e-4, 1.0 - 1e-4)

            if bbox_annotation.shape[0] == 0:
                if torch.cuda.is_available():
                    alpha_factor = torch.ones(classification.shape).cuda() * alpha
                    alpha_factor = 1. - alpha_factor
                    focal_weight = classification
                    focal_weight = alpha_factor * torch.pow(focal_weight, gamma)

                    bce = -(torch.log(1.0 - classification))

                    # cls_loss = focal_weight * torch.pow(bce, gamma)
                    cls_loss = focal_weight * bce
                    classification_losses.append(cls_loss.sum())
                    regression_losses.append(torch.tensor(0).float().cuda())

                else:
                    alpha_factor = torch.ones(classification.shape) * alpha

                    alpha_factor = 1. - alpha_factor
                    focal_weight = classification
                    focal_weight = alpha_factor * torch.pow(focal_weight, gamma)

                    bce = -(torch.log(1.0 - classification))

                    # cls_loss = focal_weight * torch.pow(bce, gamma)
                    cls_loss = focal_weight * bce
                    classification_losses.append(cls_loss.sum())
                    regression_losses.append(torch.tensor(0).float())
                continue

            IoU = calc_iou(anchors[0, :, :], bbox_annotation[:, :4]) # num_anchors x num_annotations

            IoU_max, IoU_argmax = torch.max(IoU, dim=1) # num_anchors x 1
            #Iou_max 每一行的最大值，即锚框与标注框iou的最大值，iou_argmax代表是第几个标注框

            #import pdb
            #pdb.set_trace()

            # compute the loss for classification
            targets = torch.ones(classification.shape) * -1

            if torch.cuda.is_available():
                targets = targets.cuda()

            targets[torch.lt(IoU_max, 0.4), :] = 0
            #把锚框与目标框iou低于0.4的targets值置为0
            positive_indices = torch.ge(IoU_max, 0.5)
            #iou_max 大于0.5的置为True
            num_positive_anchors = positive_indices.sum()

            assigned_annotations = bbox_annotation[IoU_argmax, :]
            ##assigned_annotations 存放与锚框 iou最大的标注框

            targets[positive_indices, :] = 0
            targets[positive_indices, assigned_annotations[positive_indices, 4].long()] = 1
            ## 把正样本所对应的锚框所对应的类别的列置为1
            # aim = torch.randint(0, 1, (1, 80))
            # tensor([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
            #          0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
            #          0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
            #          0, 0, 0, 0, 0, 0, 0, 0]])
            # aim[0, 12] = 1
            # tensor([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
            #          0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
            #          0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
            #          0, 0, 0, 0, 0, 0, 0, 0]])
            if torch.cuda.is_available():
                alpha_factor = torch.ones(targets.shape).cuda() * alpha
            else:
                alpha_factor = torch.ones(targets.shape) * alpha

            alpha_factor = torch.where(torch.eq(targets, 1.), alpha_factor, 1. - alpha_factor)
            #对应为1的位置设置为alpha_factor
            focal_weight = torch.where(torch.eq(targets, 1.), 1. - classification, classification)

            focal_weight = alpha_factor * torch.pow(focal_weight, gamma)

            bce = -(targets * torch.log(classification) + (1.0 - targets) * torch.log(1.0 - classification))
            #注意这里的log是以e为底
            #二元交叉熵损失（Binary Cross Entropy, BCE）
            # cls_loss = focal_weight * torch.pow(bce, gamma)
            # 80中为一位都使用该公式进行计算
            cls_loss = focal_weight * bce

            if torch.cuda.is_available():
                cls_loss = torch.where(torch.ne(targets, -1.0), cls_loss, torch.zeros(cls_loss.shape).cuda())
            else:
                cls_loss = torch.where(torch.ne(targets, -1.0), cls_loss, torch.zeros(cls_loss.shape))
            temp = cls_loss.sum()/torch.clamp(num_positive_anchors.float(), min=1.0)
            classification_losses.append(cls_loss.sum()/torch.clamp(num_positive_anchors.float(), min=1.0))

            # compute the loss for regression

            if positive_indices.sum() > 0:
                assigned_annotations = assigned_annotations[positive_indices, :]

                anchor_widths_pi = anchor_widths[positive_indices]
                #把正样本的anchor_widths拿出来
                #注意理解anchor_widths[positive_indices]
                anchor_heights_pi = anchor_heights[positive_indices]
                anchor_ctr_x_pi = anchor_ctr_x[positive_indices]
                anchor_ctr_y_pi = anchor_ctr_y[positive_indices]

                gt_widths  = assigned_annotations[:, 2] - assigned_annotations[:, 0]
                gt_heights = assigned_annotations[:, 3] - assigned_annotations[:, 1]
                gt_ctr_x   = assigned_annotations[:, 0] + 0.5 * gt_widths
                gt_ctr_y   = assigned_annotations[:, 1] + 0.5 * gt_heights

                # clip widths to 1
                gt_widths  = torch.clamp(gt_widths, min=1)
                gt_heights = torch.clamp(gt_heights, min=1)

                targets_dx = (gt_ctr_x - anchor_ctr_x_pi) / anchor_widths_pi
                # 整个表达式
                # targets_dx = (gt_ctr_x - anchor_ctr_x_pi) / anchor_widths_pi
                # 的意义是计算目标框和锚框在x方向上的相对位移，并将其归一化到锚框的宽度上。
                #
                # 具体地说：
                # (gt_ctr_x - anchor_ctr_x_pi)
                # 计算了真实目标框和锚框在x方向上的中心点的差值。
                # 除以anchor_widths_pi 将这个差值归一化到锚框的宽度上。
                targets_dy = (gt_ctr_y - anchor_ctr_y_pi) / anchor_heights_pi
                targets_dw = torch.log(gt_widths / anchor_widths_pi)
                targets_dh = torch.log(gt_heights / anchor_heights_pi)
                # 表达式gt_widths / anchor_widths_pi
                # 计算了真实目标框宽度和锚框宽度之间的比例。然后，对这个比例取自然对数（torch.log），得到的结果
                # targets_dw是对数空间中的相对宽度差。
                # 这样的计算通常在目标检测任务中用于计算宽度方向上的损失。对数变换有助于处理不同尺度的宽度，因为当宽度差异很大时，
                # 对数尺度上的差异会变得更加均匀。此外，对数变换还有助于模型更好地学习如何调整锚框的宽度以匹配真实的目标框。
                # 例如，如果gt_widths是100，而anchor_widths_pi是50，那么gt_widths / anchor_widths_pi将是2，而
                # torch.log(2)大约是0.693。这意味着目标框的宽度是锚框宽度的两倍，而在对数尺度上，这个差异被表示为大约0.693。

                targets = torch.stack((targets_dx, targets_dy, targets_dw, targets_dh))
                targets = targets.t()


                if torch.cuda.is_available():
                    targets = targets/torch.Tensor([[0.1, 0.1, 0.2, 0.2]]).cuda()
                else:
                    targets = targets/torch.Tensor([[0.1, 0.1, 0.2, 0.2]])
                negative_indices = 1 + (~positive_indices)

                regression_diff = torch.abs(targets - regression[positive_indices, :])

                regression_loss = torch.where(
                    torch.le(regression_diff, 1.0 / 9.0),
                    0.5 * 9.0 * torch.pow(regression_diff, 2),
                    regression_diff - 0.5 / 9.0
                )
                regression_losses.append(regression_loss.mean())
            else:
                if torch.cuda.is_available():
                    regression_losses.append(torch.tensor(0).float().cuda())
                else:
                    regression_losses.append(torch.tensor(0).float())

        return torch.stack(classification_losses).mean(dim=0, keepdim=True), torch.stack(regression_losses).mean(dim=0, keepdim=True)
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