We should add a huggingface integration to publish micro-sam and related models. See also computational-cell-analytics/medico-sam#86.
I have started with the first steps for it by:
For the next steps we need to decide how to best represent the model in HuggingFace. This is a bit more complex, because our models consist of the SAM model (prompt encoder, image encoder, mask decoder) + a custom UNETR decoder. So I am not sure if we can use the PyTorchModelHubMixin or should rather opt for a custom upload solution. See https://huggingface.co/docs/hub/models-uploading for details.
Once we have figured this out:
We should add a huggingface integration to publish micro-sam and related models. See also computational-cell-analytics/medico-sam#86.
I have started with the first steps for it by:
For the next steps we need to decide how to best represent the model in HuggingFace. This is a bit more complex, because our models consist of the SAM model (prompt encoder, image encoder, mask decoder) + a custom UNETR decoder. So I am not sure if we can use the
PyTorchModelHubMixinor should rather opt for a custom upload solution. See https://huggingface.co/docs/hub/models-uploading for details.Once we have figured this out: