In this week’s online meetup, I mentioned looking at Machine Learning-based super-resolution, that is the various ML approaches to image enhancement and upscaling.
Here’s a couple of examples, but there’s been a lot of research in the area of upscaling photos - see Papers With Code:
The use-case I had in mind was in optimising printed content, both image and textual (hopefully reducing processing time/improving quality).
I hadn’t considered the use-case for accessibility, but it sounds like an interesting area if a near realtime system could be used to help with readability of printed & scanned documents.
Knowing that text and photographs have different characteristics and that (at the very least) would benefit from different training sets to tune for either general photos or textual content, I had a look around for some existing research for text-optimised super-resolution.
Quite a few papers concentrate on image enhancement to improve Optical Character Recognition (OCR):
- Enhancing OCR Accuracy with Super Resolution
- Boosting Optical Character Recognition: A Super Resolution Approach
- Binary Document Image Super Resolution for Improved Readability and OCR Performance
Some image restoration approaches for common artifact/fault types have been found work well with text:
- CNN image restoration
- GitHub - microsoft/Bringing-Old-Photos-Back-to-Life: Bringing Old Photo Back to Life (CVPR 2020 oral)
Also, some datasets and techniques focussing on general text enhancement within photos
- GitHub - JasonBoy1/TextZoom: A super-resolution dataset of paired LR-HR scene text images (Scene text super-resolution in the wild)
- On-Device Text Image Super Resolution
- Selective Super-Resolution for Scene Text Images
- TextSR: Content-Aware Text Super-Resolution Guided by Recognition
- GitHub - huiyang865/plugnet
I haven’t done anything with this yet, but it’s an interesting area with meaningful use-cases & a lot of research to base any work on.
If anyone has any other tools, use-cases or specific issues this could help with, it would be interesting to hear your thoughts.