FWIW about a year ago I ran some experiments passing photos of Whole Foods receipts through AWS Textract. It did pretty well but error rates are still relatively high, too high to be considered reliable enough to fit into a "user-just-confirms" workflow. Some additional work on top of Textract to build receipt-format-specific models could probably get there, tho.
Many cos are successfully processing receipt photos on a commercial basis. PDFs that are machine produced are definitely processable with commodity tools, the extraction for them works quite well. I suspect sometime in the next year or two someone will put together a Jupyter notebook for doing the workflow you describe, building receipt specific models on one of the ML platforms. The basic pieces are there, it's at the level of being a solveable problem for a hobbyist. On Sat, Jan 4, 2020, 5:12 PM Philip Curtis <[email protected]> wrote: > Thanks 😁 > > Is #1 possible yet in any software technology yet? > > Thanks, > Phil > > -- > You received this message because you are subscribed to the Google Groups > "Beancount" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/beancount/e47bf4bb-62d0-4641-b0ea-edd4fe845f56%40googlegroups.com > . > -- You received this message because you are subscribed to the Google Groups "Beancount" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/beancount/CAP53tXEFDkaC69ScwPuUVJxe2atCZpGm07kT1K_A6x7XAFQtiQ%40mail.gmail.com.
