Haruya Ishikawa

Also the base config for loveDA seems to have a bug. The line for [resize pipeline](https://github.com/open-mmlab/mmsegmentation/blob/4eaa8e69191cc293b64dafe47f1f88a7d468c93c/configs/_base_/datasets/loveda.py#L10): ``` ... dict(type='Resize', img_scale=(2048, 512), ratio_range=(0.5, 2.0)), ``` shouldn't this be (1024, 1024) or...

Hi @jacobsn , Thanks for considering my project. Yes, I think having a fully differentiable version is a good idea. I haven't had time to make test scripts for checking...

Currently, I don't fully understand the situation, but if there are some scripts or error logs to work off from, I could help out. PRs are always appreciated though.

@viacheslav-m Hey thanks for adding a new function! Before I can merge your work, I'd need to test it out and write some test codes to make sure that those...

@viacheslav-m Okay, that would make understanding your code much easier since I haven't gone through papers for icosahedron transforms yet. I'd love to see some of the tests you did!

I'm trying to understand what you did. You rotated the image (call it A) and it becomes A'. And then you revert it back to the original angle (rotate by...

@helmifraser Thanks for sharing with me your workflow. Wow, that looks so much better than mine. I tried using Hugin to stitch, but can't get reliable control points since it...

Thanks for submitting an issue. Yeah, I've been struggling to make `cube2equi` -> `equi2cube` conversion as robust as possible and I've seen minor bias in my tests, but I haven't...

Oh, nvm. I misunderstood your question, you looped this 10 times. When iterating the same transform on the same data many times, it does seem the shifts increase quite a...

@qsh-zh I will document each algorithm in the future. I haven't got around to it yet :/ All algorithms (`equi2cube`, `cube2equi`, etc...) essentially do the same thing. 1. create a...