My wife had dozens, well probably over 100, handwritten recipes from a dead relative. They were pretty difficult to read. I scanned them and used mturk to have them transcribed.
Most of the work was done by one person - i think she was a woman in the Midwest, it's been like 15-years so the details are hazy. A few recipes were transcribed by people overseas but they didn't stick at it. I had to reject only one transcription.
I used mturk in some work projects too but those were boring and maybe also a little unethical (basically paying people 0.50 to give us all of their Facebook graph data, for example.)
I used Gemini to decode and transcribe an old (and well known) cursive hand written mail. I couldn't read it at all. It managed to do this in a few seconds. I am not sure if it used an already available transcription or not. However if not, it was amazing work.
Almost 2 years ago I did this with ChatGPT. It was soon after you could feed it images as input IIRC. It worked very well. I settled on AWS Textract + ChatGPT to save money and was able to get it to well under 1 cent to take an image and turn it into a recipe you could export to Paprika (and others). I never pursued it further but it was a fun little side project.
At this point I don’t think I’d do the Textract step since LLMs have gotten way better and cheaper. Also you lose some info/context when the model only gets the post-OCR data.
Most of the work was done by one person - i think she was a woman in the Midwest, it's been like 15-years so the details are hazy. A few recipes were transcribed by people overseas but they didn't stick at it. I had to reject only one transcription.
I used mturk in some work projects too but those were boring and maybe also a little unethical (basically paying people 0.50 to give us all of their Facebook graph data, for example.)