Closed Cancanneed666 closed 1 year ago
Hi, I would suggest to follow data_prepare.py in MedISeg framework to stack your own data and label to numpy file and then I think you can directly train with CTO. In case you do not use our preprocessing method, I would suggest to first modify dataloading/data preprocessing part of the code since during training, validation and inference, the data is loaded in numpy or tensor format whatever your initial files are.
Sincerely thanks for your reply and i will try this method.
Hi, i am a rookie in Medseg and i wanna train my own 2d-dataset with CTO.What should i do.Or rather, which part of the code should I start by modifying.