Spaces:
Starting
on
T4
Starting
on
T4
File size: 4,295 Bytes
a73717c |
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 |
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
from math import exp
class FocalLoss(nn.Module):
"""
copy from: https://github.com/Hsuxu/Loss_ToolBox-PyTorch/blob/master/FocalLoss/FocalLoss.py
This is a implementation of Focal Loss with smooth label cross entropy supported which is proposed in
'Focal Loss for Dense Object Detection. (https://arxiv.org/abs/1708.02002)'
Focal_Loss= -1*alpha*(1-pt)*log(pt)
:param alpha: (tensor) 3D or 4D the scalar factor for this criterion
:param gamma: (float,double) gamma > 0 reduces the relative loss for well-classified examples (p>0.5) putting more
focus on hard misclassified example
:param smooth: (float,double) smooth value when cross entropy
:param balance_index: (int) balance class index, should be specific when alpha is float
:param size_average: (bool, optional) By default, the losses are averaged over each loss element in the batch.
"""
def __init__(self, apply_nonlin=None, alpha=None, gamma=2, balance_index=0, smooth=1e-5, size_average=True):
super(FocalLoss, self).__init__()
self.apply_nonlin = apply_nonlin
self.alpha = alpha
self.gamma = gamma
self.balance_index = balance_index
self.smooth = smooth
self.size_average = size_average
if self.smooth is not None:
if self.smooth < 0 or self.smooth > 1.0:
raise ValueError('smooth value should be in [0,1]')
def forward(self, logit, target):
# logit: [B, 2, 224, 224]
# target:[B, 1, 224, 224]
if self.apply_nonlin is not None:
logit = self.apply_nonlin(logit)
# 2
num_class = logit.shape[1]
if logit.dim() > 2:
# N,C,d1,d2 -> N,C,m (m=d1*d2*...)
# [B, 2, 224*224]
logit = logit.view(logit.size(0), logit.size(1), -1)
# [B, 224*224, 2]
logit = logit.permute(0, 2, 1).contiguous()
# [B*224*224, 2]
logit = logit.view(-1, logit.size(-1))
target = torch.squeeze(target, 1)
# [B*224*224, 1]
target = target.view(-1, 1)
alpha = self.alpha
if alpha is None:
alpha = torch.ones(num_class, 1)
elif isinstance(alpha, (list, np.ndarray)):
assert len(alpha) == num_class
alpha = torch.FloatTensor(alpha).view(num_class, 1)
alpha = alpha / alpha.sum()
elif isinstance(alpha, float):
alpha = torch.ones(num_class, 1)
alpha = alpha * (1 - self.alpha)
alpha[self.balance_index] = self.alpha
else:
raise TypeError('Not support alpha type')
if alpha.device != logit.device:
alpha = alpha.to(logit.device)
# [B*224*224, 1]
idx = target.cpu().long()
# [B*224*224, 2]
one_hot_key = torch.FloatTensor(target.size(0), num_class).zero_()
one_hot_key = one_hot_key.scatter_(1, idx, 1)
if one_hot_key.device != logit.device:
one_hot_key = one_hot_key.to(logit.device)
if self.smooth:
one_hot_key = torch.clamp(
one_hot_key, self.smooth / (num_class - 1), 1.0 - self.smooth)
pt = (one_hot_key * logit).sum(1) + self.smooth
logpt = pt.log()
gamma = self.gamma
alpha = alpha[idx]
alpha = torch.squeeze(alpha)
loss = -1 * alpha * torch.pow((1 - pt), gamma) * logpt
if self.size_average:
loss = loss.mean()
return loss
class BinaryDiceLoss(nn.Module):
def __init__(self):
super(BinaryDiceLoss, self).__init__()
def forward(self, input, targets):
# 获取每个批次的大小 N
N = targets.size()[0]
# 平滑变量
smooth = 1
# 将宽高 reshape 到同一纬度
input_flat = input.view(N, -1)
targets_flat = targets.view(N, -1)
# 计算交集
intersection = input_flat * targets_flat
N_dice_eff = (2 * intersection.sum(1) + smooth) / (input_flat.sum(1) + targets_flat.sum(1) + smooth)
# 计算一个批次中平均每张图的损失
loss = 1 - N_dice_eff.sum() / N
return loss |