Train and test other datasets

[esgomezm]

I've checked the dataset.py code but I do not find the way to get my data ready to train this network.

Is there any code to configure new data as the cases in /bin/experiments/instance_segmentation/celltrackingchallenge/trainingdataset/? i.e. how are the files in each set up created? Also, the .mha files for the segmentation loss seem to be empty. Is there anything important to take into account?

Thank you very much!

[Ahmed-Noureddine-Amri]

I'm facing the issue, the code runs on the WHS dataset, but I still can't figure out how to test the model and run it on other datasets. Have you had any progress with the issue ??

[christianpayer]

For training and evaluating on other datasets, you would need to update the dataset.py files for the specific training and inference scripts. However, this could be quite difficult depending on your task... For the cell tracking and segmentation datasets this was specifically difficult to do. So my tip would be: look at the preprocessed images of the specific sub-repositories and try to preprocess your own images such that they look similar. @Ahmed-Noureddine-Amri : I created an inference script for the WHS networks. I hope this can serve as a starting point for your own experiments.

[idhamari]

@christianpayer many thanks for your effort, the framework is excellent it is well written. Probably the only thing that is less user-friendly is adapting to new datasets.