[rorylaitila]

I'm in an adjacent area, cataloging my huge collection of periodicals for my vintage ad collection (adretro.com). The biggest thing that helped me that I didn't see you mention but maybe I missed it: taking book cover photos to populate the inventory. OpenAI vision can easily extract the book, author and meta data. This speeds up the data entry considerably. I scan a whole box of periodicals and upload a zip of all the images. My software extracts the info. So for yours, if I just take photos of all the books on a shelf, it could handle the rest.

[hmkoyan]

Thank you for your comment. Currently, there are three methods of adding a book: searching (which uses the Google Books API under the hood), manual addition (which nobody wants to do — I completely understand — but is the only option for old books that cannot be found online; I have a lot of these) and scanning the ISBN (which also uses the Google Books API). I initially thought about adding the method you suggested, where you provide a picture and then extract the information, but I decided it was not the right time to implement that. So, I left it for later, when there is any need or feedback regarding it from users. From your comment, I understand that it's something that users might actually consider using.

[rorylaitila]

Yeah if there is an ISBN listed, that will be fastest. But for older books that don't have an ISBN, if there is a title on the cover or spine, OpenAI is really good at extracting that, regardless of the font or image quality. It's like a super OCR. From the title and author you can probably stub in the rest of the data, and the user can correct it if necessary.