Sejin Kim
Sejin Kim
bump Looking for a solution as well.
bump, would love this feature as well.
Will interrupt be supported for Windows eventually?
Was able to resolve this issue by changing the line to (removed `.module`): `self.network.load_state_dict(torch.load(self.opt.reload_model_path, map_location='cuda:0'))` However, this made it incompatible with the downloaded models in `training/trained_models/`
Any solution here? Same problem today with `nonechucks=0.4.2`
Faced the same issue `AttributeError: module 'numpy.typing' has no attribute 'NDArray'` Resolved by downgrading to `numpy
pinned in #radiomics channel.
https://github.com/bhklab/med-imagetools/blob/1268cbc668d2960e3dbb66955ed84b298bb1a651/imgtools/modules/segmentation.py#L192
Currently, it doesn't explicitly use any functions for alignment/positioning. This should probably be 1. tested on a heterogeneous dataset, and 2. added as a feature.
position crawling was added around version 1.5, but this isn't currently incorporated into AutoPipeline.