search

wurenkai / UltraLight-VM-UNet

[arXiv] The official code for "UltraLight VM-UNet: Parallel Vision Mamba Significantly Reduces Parameters for Skin Lesion Segmentation".
220 stars 30 forks source link

Prepare the data set #2

Open liubaoning111 opened 6 months ago

liubaoning111 commented 6 months ago

What format should your data set be formatted as input?

wurenkai commented 6 months ago

Hi! The data input in our project is the input of '.npy' format file. We have just updated the project interface with steps on how to preprocess with your own dataset, you can follow the steps to prepare your dataset. Hope this will help you.

liubaoning111 commented 6 months ago

Thank you very much for your help! What if I have multiple prospects, multiple object segmentation? How do you set up your own data?

liubaoning111 commented 6 months ago

Because I see that the data you give are single-target segmentation, there is only one prospect

wurenkai commented 6 months ago

The multi-objective segmentation task can be referred to the preparation and use of 'Synapse datasets' in VM-UNet. Also, replace 'vmunet.py' with 'UltraLight_VM-UNet.py' and change the output 'return torch.sigmoid(out0)' to 'return out0'.