Issue with the Join/Sources class

[H3pha1stos]

Hi,

I am opening this issue to report an error related to the Join class that I encountered while working through the Variational Auto-Encoder section of the DeepLearningCrashCourse book.

The deeptrack library was installed using the following command:

uv add https://github.com/DeepTrackAI/DeepTrack2.git

Below is a minimal code snippet that reproduces the error (Python 3.14):

import os
import deeptrack as dt

data_dir = os.path.join("MNIST_dataset", "mnist")
train_files = dt.sources.ImageFolder(root=os.path.join(data_dir, "train"))
test_files = dt.sources.ImageFolder(root=os.path.join(data_dir, "test"))
files = dt.sources.Join(train_files, test_files)
dt.LoadImage(files.path).update().resolve()

On my machine, this code produces the following error:

OSError: Cannot understand given URI: None.
...
TypeError: expected str, bytes or os.PathLike object, not NoneType

Could you please let me know whether this is a bug in the Join implementation or if I am misusing the API? Any guidance on how to correctly combine ImageFolder sources and load the resulting file paths would be greatly appreciated.

Thank you in advance for your help.

[giovannivolpe]

Hi,

Thanks for your feedback. It's actually a problem with the use if the API (we're still in the process of improving the documentation).

This part is wrong:

dt.LoadImage(files.path).update().resolve()

The correct approach would be:

feature = dt.LoadImage(files.path)
image = feature(train_files[0])   # or feature(test_files[0])

with 0 the image number.