Chapter09/Semantic_Segmentation_with_U_Net.ipynb

[qwrjwq]

log = Report(n_epochs)
for ex in range(n_epochs):
    N = len(trn_dl)
    for bx, data in enumerate(trn_dl):
        loss, acc = train_batch(model, data, optimizer, criterion)
        log.record(ex+(bx+1)/N, trn_loss=loss, trn_acc=acc, end='\r')

    N = len(val_dl)
    for bx, data in enumerate(val_dl):
        loss, acc = validate_batch(model, data, criterion)
        log.record(ex+(bx+1)/N, val_loss=loss, val_acc=acc, end='\r')
        
    log.report_avgs(ex+1)

RuntimeError: only batches of spatial targets supported (3D tensors) but got targets of size: : [4, 224, 224, 3]

in UnetLoss(preds, targets)
1 ce = nn.CrossEntropyLoss()
2 def UnetLoss(preds, targets):
----> 3 ce_loss = ce(preds, targets)
4 acc = (torch.max(preds, 1)[1] == targets).float().mean()
5 return ce_loss, acc

[andysingal]

log = Report(n_epochs)
for ex in range(n_epochs):
    N = len(trn_dl)
    for bx, data in enumerate(trn_dl):
        loss, acc = train_batch(model, data, optimizer, criterion)
        log.record(ex+(bx+1)/N, trn_loss=loss, trn_acc=acc, end='\r')

    N = len(val_dl)
    for bx, data in enumerate(val_dl):
        loss, acc = validate_batch(model, data, criterion)
        log.record(ex+(bx+1)/N, val_loss=loss, val_acc=acc, end='\r')
        
    log.report_avgs(ex+1)

RuntimeError: only batches of spatial targets supported (3D tensors) but got targets of size: : [4, 224, 224, 3]

in UnetLoss(preds, targets) 1 ce = nn.CrossEntropyLoss() 2 def UnetLoss(preds, targets): ----> 3 ce_loss = ce(preds, targets) 4 acc = (torch.max(preds, 1)[1] == targets).float().mean() 5 return ce_loss, acc

I am also getting error:

RuntimeError                              Traceback (most recent call last)
[<ipython-input-11-74e7caf8d75b>](https://localhost:8080/#) in <cell line: 2>()
      3     N = len(trn_dl)
      4     for bx, data in enumerate(trn_dl):
----> 5         loss, acc = train_batch(model, data, optimizer, criterion)
      6         log.record(ex+(bx+1)/N, trn_loss=loss, trn_acc=acc, end='\r')
      7 

5 frames
[/usr/local/lib/python3.10/dist-packages/torch/nn/functional.py](https://localhost:8080/#) in cross_entropy(input, target, weight, size_average, ignore_index, reduce, reduction, label_smoothing)
   3057     if size_average is not None or reduce is not None:
   3058         reduction = _Reduction.legacy_get_string(size_average, reduce)
-> 3059     return torch._C._nn.cross_entropy_loss(input, target, weight, _Reduction.get_enum(reduction), ignore_index, label_smoothing)
   3060 
   3061 

RuntimeError: only batches of spatial targets supported (3D tensors) but got targets of dimension: 4

[WalidGharianiEAGLE]

• Issue: The original read function from torch_snippets was causing issues when reading the mask images from SegData.
• Solution: I replaced read with cv2.imread.

class SegData(Dataset):
    def __init__(self, split):
        self.items = stems(f'./dataset1/dataset1/images_prepped_{split}')
        self.split = split
    def __len__(self):
        return len(self.items)
    def __getitem__(self, ix):
        image = read(f'./dataset1/images_prepped_{self.split}/{self.items[ix]}.png', 1)
        image = cv2.resize(image, (224,224))
        # read(f'./dataset1/annotations_prepped_{self.split}/{self.items[ix]}.png')
        mask = cv2.imread(f'./dataset1/annotations_prepped_{self.split}/{self.items[ix]}.png', cv2.IMREAD_GRAYSCALE) 
        mask = cv2.resize(mask, (224,224))
        return image, mask
    def choose(self): return self[randint(len(self))]
    def collate_fn(self, batch):
        ims, masks = list(zip(*batch))
        ims = torch.cat([tfms(im.copy()/255.)[None] for im in ims]).float().to(device)
        ce_masks = torch.cat([torch.Tensor(mask[None]) for mask in masks]).long().to(device)
        return ims, ce_masks