Sébastien Blaise

Thanks. After a first try, it does not seem straightforward to build the project on VS2013: C++17 features, and CoverageExt project is incompatible with the VS2013 version and cannot load....

You need to setup mmsegmentation / mmdetection or other "mm" appropriate framework, that is where tool/test.py and tool/train.py are. You can have a look at #6

Thanks for your reply. By updating the dataloader and some parameters in the config file for model/preprocessing, I can run rvsa-l on on a semantic segmentation task with 13-channel input...

Here: https://github.com/microsoft/Swin-Transformer/blob/main/MODELHUB.md

It still fails with numba 0.57.0

Still affecting numba-0.60.0

I would say you need interpolation since your image has not the same resolution as the trained model (higher or lower). Some details here: https://discuss.huggingface.co/t/fine-tuning-image-transformer-on-higher-resolution/22623/6