Why is handwriting recognition so bad? One would think that if we can build a machine to recognize cat pictures, recognizing a handwritten "A" ought to be trivial.
My handwriting was described by a professor at college as "dog's dribble" and has only degraded as I spent most of my writing time typing (or painting script!) for the last many years. I've been super surprised to find that doing a search in OneNote will find some of my marginal notes, in part because I knew it was OCR-ing but just imagined it would only address formal fonts/body text, and in part because that writing is really poor.
OneNote is the best software I've used from MS I think, VS Code made me wonder if I needed to reassess MS, but OneNote is super-effective. I moved 2 years ago to a paper-free work process (by choice when I started a new job) and OneNote has been central to that. It has flaws a-plenty (not too many bugs that affect me), but it just really fits for me.
Short version, MS's OCR is getting there. However, it's not 'recognise "A"' that's the issue, I'm a word-based handwriter, it's the general shape of the word I record in my writing when it's at its worst; that A might be more like a capital lambda with a floating platform in the general vicinity.
Why is handwriting recognition so bad? One would think that if we can build a machine to recognize cat pictures, recognizing a handwritten "A" ought to be trivial.