How to train with multiple mask labels

[ncp-nice]

The model only supports the mask in the one-hot format, if I have a mask of labels, should I extract each label separately according to the pixels and then train it? It's also too much trouble！

[ncp-nice]

I would like to ask how the author trained and tested the Synapse multi-organ CT dataset, and did he have to convert each multi-label label to the onehot format one by one? thanks！

[fengyasi]

Are you asking about labeling multiple tags on a single image?

[ncp-nice]

Are you asking about labeling multiple tags on a single image?

Doesn't the SAMMED2d model require that the labels corresponding to the image data of train and test be in one-hot format? Therefore, when encountering medical images in 3d format, do you have to slice them first and then classify them separately according to each label before they can be put into the SAMMED2d model for training? But for example, like the Synapse multi-organ CT dataset, I slice it, but it won't separate the label into a one-hot format

[ncp-nice]

Are you asking about labeling multiple tags on a single image?

Yes