nitr098 / awesome-u-net Goto Github PK

Which file needs to be modified for the path to the dataset? Take UNet as an example, I changed the dataset path in line 41 of isic.py（And config file）, but it still reports an error, showing "FileNotFoundError: [Errno 2] No such file or directory: 'D:/Dataset/ISIC2018/np/X_tr_224x224.npy'"

Hello, thank you for doing such a excellent job！I'm a little confused about this data set and metrics.
Can you help me answer them?
First, as far as I know, the data set of ISIC Task 2018 officially has validation set and test set. May I ask why the training set should be separately divided into training set, validation set and test set? Instead of using the official validation set and test set?
Second question, in this paper, do the relevant indicators in Table 2 (a) refer to the validation set or the test set? I have downloaded your pre-training weights and found that they do not match the results in the paper.

Hi
may I ask which version of torchvision did you use in your code? I use 0.6.0 and some features of this have changed, so I do not know which version to use. Thank you very much!

In the preparation of the data set, in the prepare_segpc.ipynb file, the specific settings of the sim_resize and resize_pad functions are not given. When I run it, I get an error NameError: name 'sim_resize' is not defined And NameError: name 'resize_pad' is not defined.Please tell me how to solve it, thank you

Hello, may I ask which version of torchmetrics did you use in your code? I just run the code in the simply unet with isic dataset. I used torchmetrics=='0.11.1', but when I runned "torchmetrics.F1Score()", it showed "new() missing 1 required positional argument: 'task'". Then, I added "torchmetrics.F1Score(task="multiclass", num_classes=2)", so it has successed. But, next, when I trained, it showed "size mismatch for de_4.weight: copying a param with shape torch.Size([1, 32, 1, 1]) from checkpoint, the shape in current model is torch.Size([2, 32, 1, 1]).". I don't know how to do it. So, please help me. I'm looking forward to receiving your reply.

Hi,Thanks for your tremendous and excellent work！！！
I found that I couldn't download the pre-trained weights. Can you upload it again?

Thank you for your great job.
It would be highly beneficial if the code for the preparation of the Synapse Dataset could be included in the repository as well.

Thanks

Hello, thank you for doing such a excellent job！I'm a little confused about this data set and metrics.
Can you help me answer them?