RuntimeError: 1only batches of spatial targets supported (3D tensors) but got targets of size: : [2, 512, 512, 3]

[Moonquakes0]

您好，我尝试用您的代码训练的时候出现了如下问题。似乎是在计算loss的时候target应该是三维，但是debug进去发现label却还是四维，不知道该如何解决呢

Traceback (most recent call last):
File "/data/yuanyifeng/Bi-SRNet-main/train_SCD.py", line 238, in
main()
File "/data/yuanyifeng/Bi-SRNet-main/train_SCD.py", line 65, in main
train(train_loader, net, criterion, optimizer, scheduler, args, val_loader)
File "/data/yuanyifeng/Bi-SRNet-main/train_SCD.py", line 105, in train
loss_seg = criterion(outputs_A, labels_A) * 0.5 + criterion(outputs_B, labels_B) * 0.5
File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "/data/yuanyifeng/Bi-SRNet-main/utils/loss.py", line 13, in forward
return self.nll_loss(F.log_softmax(inputs, dim=1), targets)
File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/modules/loss.py", line 213, in forward
return F.nll_loss(input, target, weight=self.weight, ignore_index=self.ignore_index, reduction=self.reduction)
File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/functional.py", line 2266, in nll_loss
ret = torch._C._nn.nll_loss2d(input, target, weight, _Reduction.get_enum(reduction), ignore_index)
RuntimeError: 1only batches of spatial targets supported (3D tensors) but got targets of size: : [2, 512, 512, 3]

[ggsDing]

Please check if you have used my codes to load the data. Also please check if you have loaded the RGB labels or the index labels.

[ggsDing]

If you prefer to use RGB labels, you can use the Colorls2Index function in RS_ST.py

[Moonquakes0]

@ggsDing Thanks for your reply, I will try again.

[Evan1127-Lin]

@ggsDing Thanks for your reply, I will try again.

Hello, I also encounter the same problem as you met. Have you succeeded addressing this issue?

[Evan1127-Lin]

@ggsDing Excuse me, I have tried adding the following codes:
labels_A = RS.Colorls2Index(labels_A)
labels_B = RS.Colorls2Index(labels_B)
after the code part:
if args['gpu']:
imgs_A = imgs_A.cuda().float()
imgs_B = imgs_B.cuda().float()
labels_bn = (labels_A>0).unsqueeze(1).cuda().float()
labels_A = labels_A.cuda().long()
labels_B = labels_B.cuda().long()
in train_SCD.py. But another problem arises when running this file as shown below:
File "/home/lyz/code/supervised/Bi-SRNet/Bi-SRNet-main/train_SCD.py", line 100, in train
labels_A = RS.Colorls2Index(labels_A)
File "/home/lyz/code/supervised/Bi-SRNet/Bi-SRNet-main/datasets/RS_ST.py", line 33, in Colorls2Index
IndexMap = Color2Index(data)
File "/home/lyz/code/supervised/Bi-SRNet/Bi-SRNet-main/datasets/RS_ST.py", line 38, in Color2Index
data = ColorLabel.astype(np.int32)
AttributeError: 'Tensor' object has no attribute 'astype'
Could you please give me some specific help? Thank you in advance!

[Evan1127-Lin]

@ggsDing I have succeeded solving the issue mentioned above. The solution is modifying "RS_ST.py" file by adding 2 code lines at the last of function "read_RSimages":
labels_A = Colorls2Index(labels_A)
labels_B = Colorls2Index(labels_B)
Thank you all the same!

[DA-II]

之前粘贴错了位置 感谢Evan1127-Lin提出的方法 注意一定要粘贴到read_RSimages的尾部部不是read_RSimages 内 for循环尾部 Colorls2Index这个函数的作用是把rgb的labels（512，512，3）转化为以类为标签（512，512）型数据

[ggsDing]

@ggsDing I have succeeded solving the issue mentioned above. The solution is modifying "RS_ST.py" file by adding 2 code lines at the last of function "read_RSimages": labels_A = Colorls2Index(labels_A) labels_B = Colorls2Index(labels_B) Thank you all the same!

Glad that you found a solution. I have added codes for rgb labels in the updated RS_ST.py lines 93-94.

[Gchang9]

您好，我尝试用您的代码训练的时候出现了如下问题。似乎是在计算loss的时候target应该是三维，但是debug进去发现label却还是四维，不知道该如何解决呢

Traceback (most recent call last): File "/data/yuanyifeng/Bi-SRNet-main/train_SCD.py", line 238, in main() File "/data/yuanyifeng/Bi-SRNet-main/train_SCD.py", line 65, in main train(train_loader, net, criterion, optimizer, scheduler, args, val_loader) File "/data/yuanyifeng/Bi-SRNet-main/train_SCD.py", line 105, in train loss_seg = criterion(outputs_A, labels_A) * 0.5 + criterion(outputs_B, labels_B) * 0.5 File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/data/yuanyifeng/Bi-SRNet-main/utils/loss.py", line 13, in forward return self.nll_loss(F.log_softmax(inputs, dim=1), targets) File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/modules/loss.py", line 213, in forward return F.nll_loss(input, target, weight=self.weight, ignore_index=self.ignore_index, reduction=self.reduction) File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/functional.py", line 2266, in nll_loss ret = torch._C._nn.nll_loss2d(input, target, weight, _Reduction.get_enum(reduction), ignore_index) RuntimeError: 1only batches of spatial targets supported (3D tensors) but got targets of size: : [2, 512, 512, 3]

我也遇到了这个问题，请问您解决了吗？如果解决了烦请赐教！我用的是不是RGB的标签。

[ggsDing]

您好，我尝试用您的代码训练的时候出现了如下问题。似乎是在计算loss的时候target应该是三维，但是debug进去发现label却还是四维，不知道该如何解决呢
Traceback (most recent call last): File "/data/yuanyifeng/Bi-SRNet-main/train_SCD.py", line 238, in main() File "/data/yuanyifeng/Bi-SRNet-main/train_SCD.py", line 65, in main train(train_loader, net, criterion, optimizer, scheduler, args, val_loader) File "/data/yuanyifeng/Bi-SRNet-main/train_SCD.py", line 105, in train loss_seg = criterion(outputs_A, labels_A) * 0.5 + criterion(outputs_B, labels_B) * 0.5 File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/data/yuanyifeng/Bi-SRNet-main/utils/loss.py", line 13, in forward return self.nll_loss(F.log_softmax(inputs, dim=1), targets) File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/modules/loss.py", line 213, in forward return F.nll_loss(input, target, weight=self.weight, ignore_index=self.ignore_index, reduction=self.reduction) File "/home/yuanyifeng/anaconda3/envs/ASN3/lib/python3.7/site-packages/torch/nn/functional.py", line 2266, in nll_loss ret = torch._C._nn.nll_loss2d(input, target, weight, _Reduction.get_enum(reduction), ignore_index) RuntimeError: 1only batches of spatial targets supported (3D tensors) but got targets of size: : [2, 512, 512, 3]

我也遇到了这个问题，请问您解决了吗？如果解决了烦请赐教！我用的是不是RGB的标签。

For RGB labels please uncomment lines 93-94 in RS_ST.py