-
Notifications
You must be signed in to change notification settings - Fork 0
/
losses.py
188 lines (162 loc) · 6.92 KB
/
losses.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
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.autograd import Variable
class CharbonnierLoss(nn.Module):
"""Charbonnier Loss (L1)"""
def __init__(self, eps=1e-3):
super(CharbonnierLoss, self).__init__()
self.eps = eps
def forward(self, x, y):
diff = x - y
# loss = torch.sum(torch.sqrt(diff * diff + self.eps))
loss = torch.mean(torch.sqrt((diff * diff) + (self.eps*self.eps)))
return loss
# class EdgeLoss(nn.Module):
# def __init__(self):
# super(EdgeLoss, self).__init__()
# k = torch.Tensor([[.05, .25, .4, .25, .05]])
# self.kernel = torch.matmul(k.t(),k).unsqueeze(0).repeat(3,1,1,1)
# if torch.cuda.is_available():
# self.kernel = self.kernel.cuda()
# self.loss = CharbonnierLoss()
#
# def conv_gauss(self, img):
# n_channels, _, kw, kh = self.kernel.shape
# img = F.pad(img, (kw//2, kh//2, kw//2, kh//2), mode='replicate')
# return F.conv2d(img, self.kernel, groups=n_channels)
#
# def laplacian_kernel(self, current):
# filtered = self.conv_gauss(current) # filter
# down = filtered[:,:,::2,::2] # downsample
# new_filter = torch.zeros_like(filtered)
# new_filter[:,:,::2,::2] = down*4 # upsample
# filtered = self.conv_gauss(new_filter) # filter
# diff = current - filtered
# return diff
#
# def forward(self, x, y):
# loss = self.loss(self.laplacian_kernel(x), self.laplacian_kernel(y))
# return loss
class CEL(nn.Module):
def __init__(self):
super(CEL, self).__init__()
#print("You are using `CEL`!")
self.eps = 1e-6
def forward(self, pred, target):
#pred = pred.sigmoid()
intersection = pred * target
numerator = (pred - intersection).sum() + (target - intersection).sum()
denominator = pred.sum() + target.sum()
return numerator / (denominator + self.eps)
def iou_loss(pred, mask):
#pred = torch.sigmoid(pred)
inter = (pred*mask).sum(dim=(2,3))
union = (pred+mask).sum(dim=(2,3))
iou = 1-(inter+1)/(union-inter+1)
return iou.mean()
############################################ edge loss #################################################
def cross_entropy(logits, labels):
return torch.mean((1 - labels) * logits + torch.log(1 + torch.exp(-logits)))
class EdgeLoss(nn.Module):
def __init__(self):
super().__init__()
laplace = torch.FloatTensor([[-1,-1,-1,],[-1,8,-1],[-1,-1,-1]]).view([1,1,3,3])
# filter shape in Pytorch: out_channel, in_channel, height, width
self.laplace = nn.Parameter(data=laplace, requires_grad=False).cuda()
def torchLaplace(self, x):
edge = F.conv2d(x, self.laplace, padding=1)
edge = torch.abs(torch.tanh(edge))
return edge
def forward(self, y_pred, y_true, mode=None):
y_true_edge = self.torchLaplace(y_true)
y_pred_edge = self.torchLaplace(y_pred)
edge_loss = cross_entropy(y_pred_edge, y_true_edge)
return edge_loss
######################################## Focal_loss #############################################
class FocalLossV1(nn.Module):
def __init__(self,
alpha=0.25,
gamma=2,
reduction='mean',):
super(FocalLossV1, self).__init__()
self.alpha = alpha
self.gamma = gamma
self.reduction = reduction
self.crit = nn.BCEWithLogitsLoss(reduction='none')
self.class_num = 1
def _one_hot_encoder(self, input_tensor):
tensor_list = []
for i in range(self.class_num):
temp_prob = input_tensor == i # * torch.ones_like(input_tensor)
tensor_list.append(temp_prob.unsqueeze(1))
output_tensor = torch.cat(tensor_list, dim=1)
return output_tensor.float()
def forward(self, logits, label):
# '''
# Usage is same as nn.BCEWithLogits:
# >>> criteria = FocalLossV1()
# >>> logits = torch.randn(8, 19, 384, 384)
# >>> lbs = torch.randint(0, 2, (8, 19, 384, 384)).float()
# >>> loss = criteria(logits, lbs)
# '''
#probs = torch.sigmoid(logits)
probs = logits
label = self._one_hot_encoder(label)
coeff = torch.abs(label - probs).pow(self.gamma).neg()
log_probs = torch.where(logits >= 0,
F.softplus(logits, -1, 50),
logits - F.softplus(logits, 1, 50))
log_1_probs = torch.where(logits >= 0,
-logits + F.softplus(logits, -1, 50),
-F.softplus(logits, 1, 50))
loss = label * self.alpha * log_probs + (1. - label) * (1. - self.alpha) * log_1_probs
loss = loss * coeff
if self.reduction == 'mean':
loss = loss.mean()
if self.reduction == 'sum':
loss = loss.sum()
return loss
############################### Floss ############################################################
class FLoss(nn.Module):
def __init__(self, beta=0.3, log_like=True):
super(FLoss, self).__init__()
self.beta = beta
self.log_like = log_like
def forward(self, prediction, target):
EPS = 1e-10
N = prediction.size(0)
TP = (prediction * target).view(N, -1).sum(dim=1)
H = self.beta * target.view(N, -1).sum(dim=1) + prediction.view(N, -1).sum(dim=1)
fmeasure = (1 + self.beta) * TP / (H + EPS)
if self.log_like:
floss = -torch.log(fmeasure)
else:
floss = (1 - fmeasure)
return floss
####################################################
##### This is focal loss class for multi class #####
####################################################
# I refered https://github.com/c0nn3r/RetinaNet/blob/master/focal_loss.py
class FocalLoss2d(nn.modules.loss._WeightedLoss):
def __init__(self, gamma=2, weight=None, size_average=None, ignore_index=-100,
reduce=None, reduction='mean', balance_param=0.25):
super(FocalLoss2d, self).__init__(weight, size_average, reduce, reduction)
self.gamma = gamma
self.weight = weight
self.size_average = size_average
self.ignore_index = ignore_index
self.balance_param = balance_param
def forward(self, input, target):
# inputs and targets are assumed to be BatchxClasses
assert len(input.shape) == len(target.shape)
assert input.size(0) == target.size(0)
assert input.size(1) == target.size(1)
weight = Variable(self.weight)
# compute the negative likelyhood
logpt = - F.binary_cross_entropy_with_logits(input, target, pos_weight=weight, reduction=self.reduction)
pt = torch.exp(logpt)
# compute the loss
focal_loss = -((1 - pt) ** self.gamma) * logpt
balanced_focal_loss = self.balance_param * focal_loss
return balanced_focal_loss