improve qr detection #17
more exams are slipping through with undetected qr codes than i might like. i had played around with contrast, blurring, etc, and i hadn't found anything all that reliable. but maybe we could do something like what's described here: https://kdmurray.id.au/post/2022-03-21_decode-qrcodes/
i think actually the big size of the images hurts us. maybe we want to downsample first? that might also improve the quality of the output pdfs, now that i think of it (they're kind of terrible, maybe since we're letting the transform do the downscaling for us?).
I have an attempt at this in
8f47a041a4. After some quick testing, it does seem to work better (but it can be really slow). Maybe we also need an option to toggle slow/med/fast.
Deleting a branch is permanent. Although the deleted branch may exist for a short time before cleaning up, in most cases it CANNOT be undone. Continue?