--- a
+++ b/semseg_train/loss.py
@@ -0,0 +1,94 @@
+import torch
+from torch import nn
+import torch.nn.functional as F
+
+def soft_jaccard(outputs, targets):
+    eps = 1e-15
+    jaccard_target = (targets == 1).float()
+    jaccard_output = torch.sigmoid(outputs)
+
+    intersection = (jaccard_output * jaccard_target).sum()
+    union = jaccard_output.sum() + jaccard_target.sum()
+    return intersection / (union - intersection + eps)
+
+def structure_loss(pred, mask):
+    #重み付きloss
+    #ref:https://github.com/DengPingFan/PraNet
+    weit = 1 + 5*torch.abs(F.avg_pool2d(mask, kernel_size=31, stride=1, padding=15) - mask)
+    wbce = F.binary_cross_entropy_with_logits(pred, mask, reduction='none')
+    wbce = (weit*wbce).sum(dim=(2, 3)) / weit.sum(dim=(2, 3))
+
+    pred = torch.sigmoid(pred)
+    inter = ((pred * mask)*weit).sum(dim=(2, 3))
+    union = ((pred + mask)*weit).sum(dim=(2, 3))
+    wiou = 1 - (inter + 1)/(union - inter+1)
+    return (wbce + wiou).mean()
+
+class LossBinary:
+    """
+    Loss defined as BCE - log(soft_jaccard)
+
+    Vladimir Iglovikov, Sergey Mushinskiy, Vladimir Osin,
+    Satellite Imagery Feature Detection using Deep Convolutional Neural Network: A Kaggle Competition
+    arXiv:1706.06169
+    """
+
+    def __init__(self, jaccard_weight=0):
+        self.nll_loss = nn.BCEWithLogitsLoss()#the sum of the output will be divided by the number of elements(sum data) in the output
+        self.jaccard_weight = jaccard_weight
+
+    def __call__(self, outputs, targets):
+        loss = (1 - self.jaccard_weight) * self.nll_loss(outputs, targets)
+
+        if self.jaccard_weight:
+            loss += self.jaccard_weight * (1 - soft_jaccard(outputs, targets))
+        return loss
+
+class LossBCE:
+    """
+    Loss defined as BCE - log(soft_jaccard)
+
+    Vladimir Iglovikov, Sergey Mushinskiy, Vladimir Osin,
+    Satellite Imagery Feature Detection using Deep Convolutional Neural Network: A Kaggle Competition
+    arXiv:1706.06169
+    """
+
+    def __init__(self, jaccard_weight=0):
+        self.nll_loss = nn.BCEWithLogitsLoss(reduction='sum')#1ピクセル当たりのBCE（実数1つを返す）、バッチ数で割り画像サイズ縦*横で割る
+        self.jaccard_weight = jaccard_weight
+
+    def __call__(self, outputs, targets):
+        loss = (1 - self.jaccard_weight) * self.nll_loss(outputs, targets)
+        return loss
+
+class LossJaccard:
+    """
+    Loss defined as BCE - log(soft_jaccard)
+
+    Vladimir Iglovikov, Sergey Mushinskiy, Vladimir Osin,
+    Satellite Imagery Feature Detection using Deep Convolutional Neural Network: A Kaggle Competition
+    arXiv:1706.06169
+    """
+
+    def __init__(self, jaccard_weight=0):
+        self.jaccard_weight = jaccard_weight
+
+    def __call__(self, outputs, targets):
+        loss = self.jaccard_weight * (1 - soft_jaccard(outputs, targets))
+        return loss
+
+class LossWeightBCE:
+    """
+    Loss defined as WeightBCE
+
+    Deng-Ping Fan, Ge-Peng Ji, Tao Zhou, Geng Chen, Huazhu Fu, Jianbing Shen, and Ling Shao,
+    PraNet: Parallel Reverse Attention Network for Polyp Segmentation
+    """
+
+    #def __init__(self, jaccard_weight=0):
+        #self.nll_loss = nn.BCEWithLogitsLoss()
+        #self.jaccard_weight = jaccard_weight
+
+    def __call__(self, outputs, targets):
+        loss = structure_loss(outputs, targets)
+        return loss
\ No newline at end of file