mberkay0 / pretrained-backbones-unet Goto Github PK

I am trying to train using the efficientnet_b0 backbone using an image size of 64x48, but get the following error:

  File "/home/seg/bbunet/train.py", line 85, in <module>
    trainer.fit(train_loader, val_loader)
  File "/home/seg/bbunet/backbones_unet/utils/trainer.py", line 82, in fit
    self._train_one_epoch(train_loader, epoch)
  File "/home/seg/bbunet/backbones_unet/utils/trainer.py", line 111, in _train_one_epoch
    preds = self.model(images)
  File "/home/miniconda3/envs/pt/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
    return forward_call(*args, **kwargs)
  File "/home/seg/bbunet/backbones_unet/model/unet.py", line 174, in forward
    x = self.decoder(x)
  File "/home/miniconda3/envs/pt/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
    return forward_call(*args, **kwargs)
  File "/home/seg/bbunet/backbones_unet/model/unet.py", line 334, in forward
    x = b(x, skip)
  File "/home/miniconda3/envs/pt/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
    return forward_call(*args, **kwargs)
  File "/home/seg/bbunet/backbones_unet/model/unet.py", line 276, in forward
    x = torch.cat([x, skip], dim=1)
RuntimeError: Sizes of tensors must match except in dimension 1. Expected size 4 but got size 3 for tensor number 1 in the list.

I tried with and without pretrained model but got the same error each time.

Hello,

I am trying to use your framework to perform mask extraction on images I have with the following code :

import os

import torchvision.transforms as transforms
from backbones_unet.model.unet import Unet
from PIL import Image

PATH="path_to_folder" # Replaced for confidentiality reasons 
CELLS_PATH = PATH + "cells/"

# Model init
model = Unet(backbone='xception71', in_channels=3, num_classes=1)
transform = transforms.Compose([transforms.ToTensor()])

# Data processing
for filename in os.listdir(CELLS_PATH):
    image = Image.open(CELLS_PATH + filename)
    tensor = transform(image).unsqueeze(0)
    print(tensor.shape)
    mask = model.predict(tensor)
    print(mask)

For the moment I just print masks to check what they look like before saving them. But I get the following error when trying to run the code :

Traceback (most recent call last):
File "hidden/mask.py", line 19, in
mask = model.predict(tensor)
File "env/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "env/lib/python3.10/site-packages/backbones_unet/model/unet.py", line 160, in predict
x = self.forward(x)
File "env/lib/python3.10/site-packages/backbones_unet/model/unet.py", line 142, in forward
x = self.decoder(x)
File "env/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "env/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "env/lib/python3.10/site-packages/backbones_unet/model/unet.py", line 304, in forward
x = b(x, skip)
File "env/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "env/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "env/lib/python3.10/site-packages/backbones_unet/model/unet.py", line 246, in forward
x = torch.cat([x, skip], dim=1)
RuntimeError: Sizes of tensors must match except in dimension 1. Expected size 14 but got size 13 for tensor number 1 in the list.

And when printing dimensions of my input tensor as well as x and skip which are concatenated, I get in the same order :

torch.Size([1, 3, 207, 204])
torch.Size([1, 2048, 14, 14])
torch.Size([1, 1024, 13, 13])

Would you know where this come from / how to fix it ?

Do you have a working test script for these models? I just need to get prediction masks from a test set, I don't need metrics.

Hi, when I am trying to start the training I am getting the following error:

NameError                                 Traceback (most recent call last)
/tmp/ipykernel_21140/3294596851.py in <module>
     14 val_dataset = SemanticSegmentationDataset(val_img_path, val_mask_path)
     15 
---> 16 train_loader = DataLoader(train_dataset, batch_size=2)
     17 val_loader = DataLoader(val_dataset, batch_size=2)
     18 

NameError: name 'DataLoader' is not defined

Do you have any suggestions to how to solve it?

Thanks!

Thanks for creating this library, I was trying to implement the same, coz segmentation models pytorch library doesn't really provide the convnext as encoder.

I hope you implement the saving functionality soon..

Thank you.

Hello, thank you for the great library.

There is a line in the DiceLoss class:

if not self.from_logits: y_pred = F.sigmoid(y_pred)

I am not sure, but from the description, I think it is meant to apply the sigmoid if the self.from_logits is True, so there should be just "if", instead of "if not".

Hello, I am using Unet decoder and HRNet encoder model but i realize that the training loss is not decreasing, instead it seems to remain the same. Also the training logs appear in separate lines which is a problem becuase 1 epoch take around 200 lines and not gets output in one line.

I have trained it for 10 epochs but the loss keeps jumping between 0.99 and 1

In trainer.py (line 101) there is a missing argument. Here's how I define optimiser and lr_scheduler in train.py:

optimizer = torch.optim.Adam(params, lr=0.001, weight_decay=0.0001)
scheduler = lr_scheduler.ReduceLROnPlateau(optimizer, factor=0.1, patience=6, verbose=1, min_lr=0.000001)
trainer = Trainer(
    model,
    criterion=BCEDiceLoss(),
    optimizer=optimizer,
    lr_scheduler=scheduler,   
    epochs=500
)

That throws the following error:

File "/home/seg/backbones_unet/utils/trainer.py", line 127, in _train_one_epoch
    self.lr_scheduler.step() # this was originally: self.lr_scheduler.step()
TypeError: step() missing 1 required positional argument: 'metrics'

If I then set line 101 to self.lr_scheduler.step(loss) that seems to fix the error. However, when I start training I get this:

Training Model on 500 epochs:   0%|                            | 0/500 [00:00<?, ?it/s
Epoch 00032: reducing learning rate of group 0 to 1.0000e-04.  | 31/800 [00:03<00:54, 14.03 training-batch/s, loss=1.1]
Epoch 00054: reducing learning rate of group 0 to 1.0000e-05.  | 53/800 [00:04<00:51, 14.38 training-batch/s, loss=1.09]
Epoch 00062: reducing learning rate of group 0 to 1.0000e-06.  | 61/800 [00:05<00:54, 13.45 training-batch/s, loss=1.06]
^Epoch 1:  13%|██████████████████                              | 104/800 [00:08<00:58, 11.84 training-batch/s, loss=1.05]

I haven't seen that before when training models with code from other repos. If that is normal, then all is OK, I just wanted to report the missing argument error in trainer.py.