Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Automating receipt processing with deep learning

nanonets.com

Post reply on HN
Print view View original

11–20 of 35 posts

Re: Automating receipt processing with deep learning

#11

Post by BrandiATMuhkuh »

amelius wrote: post #10
axiosgunnar wrote: post #3

What is the point of spamming [1] HN with this low-effort content marketing for Nanonets that barely scratches the surface of the topic but instead is full of obnoxious calls to action? [1]: https://i.imgur.com/eIhwtjo.png

> Need to digitize documents, receipts or invoices but too lazy to code? Head over to Nanonets and build OCR models for free! I think that once you go down the path of OCR'ing your own invoices, you better start a company around it.

That's correct. I'm follow this topic since about 10 years now. It seems to be an important issues. But super hard problem. I could not find any close to relatable solution.

Building a company that does only exactly one things, offer an API that takes a receipt as input and returns key/value, is enough.

Re: Automating receipt processing with deep learning

#12

Post by henrikschroder »

tastyminerals wrote: post #7

It's a nice overview article for anyone interested in the topic of IE from financial documents. However, for industrial level solutions Tesseract does not cut it. Abbyy is the best OCR engine on the market currently. Receipt IE works just fine with rules supported by a small BiLSTM as fallback just because receipts do not contain a lot of text. With invoices this approach is suboptimal. On a general note no DL approa…

> On a general note no DL approach will give you fast and high enough results just because any advanced network would be too slow and too generic.

Yeah, but see, I have this shiny hammer, and if I squint just right, everything looks like a nail!

The seasonal trends in the programmer world get kinda tiring after a while, and the people peddling the latest hot new thing equally so, such as this thinly veiled promo piece for nanonets.

In science, there's this concept of falsifiability. If a theory can't be disproven, it's automatically false. The same goes for technological evangelism. If no-one knows what a piece of tech is bad at, how the hell would you know if it's any good at anything, really? There are no panaceas, no one-tool-to-rule-them-all, no single piece of tech that will usher in a new golden age for programmerkind. They're just tools in a toolbox. Know what each tool is good at. Know what each tool is bad at. Don't forget your old tools, just because the newest tool is still very shiny.

Platitudes, I know, but still.

Re: Automating receipt processing with deep learning

#14

Post by feistypharit »

gatestone wrote: post #13

So what is the state of art? Is there a decent software available, that actually works? An app that a small company could use easily, say, for employee expense reports with reliable data parsing and automatic expense/tax calculations?

I've been using waveapps.com. their receipt processor allows me to forward any I get in email via pdf. They have an app to upload a pic as well. Have yet to have an issue. Even the Costco ones that have a line through them. It's been great. And it's "free".

Re: Automating receipt processing with deep learning

#15

Post by huffmsa »

axiosgunnar wrote: post #3

What is the point of spamming [1] HN with this low-effort content marketing for Nanonets that barely scratches the surface of the topic but instead is full of obnoxious calls to action? [1]: https://i.imgur.com/eIhwtjo.png

I found it useful, if only for the idea.

My wife has gotten very serious about optimizing the last hay penny out of our expenses, and I'd much rather build her our own receipt ingestion process than use a 3rd party who is reselling our data.

Re: Automating receipt processing with deep learning

#16

Post by TrackerFF »

amelius wrote: post #10
axiosgunnar wrote: post #3

What is the point of spamming [1] HN with this low-effort content marketing for Nanonets that barely scratches the surface of the topic but instead is full of obnoxious calls to action? [1]: https://i.imgur.com/eIhwtjo.png

> Need to digitize documents, receipts or invoices but too lazy to code? Head over to Nanonets and build OCR models for free! I think that once you go down the path of OCR'ing your own invoices, you better start a company around it.

Gonna be honest, I think it's a dead-end. The solution IMO will come from the supply side; Easier to get receipts digitally from the store / vendor / seller, than to spend all this effort on converting physical receipts to digital.

Sure - there's good retro usage, I'd rather push for sellers to offer you digital receipts.

In fact, I already get that from my local supermarket. My bank card is registered to their app, so every time I swipe my card or scan my app (when paying with cash), I get a receipt on the app, which I can export.

I remember back in college, over 10 years ago, this was a very hot topic. Receipt management was one of those entrepreneur ideas that would always pop up.

Re: Automating receipt processing with deep learning

#18

Post by pheme1 »

axiosgunnar wrote: post #3

What is the point of spamming [1] HN with this low-effort content marketing for Nanonets that barely scratches the surface of the topic but instead is full of obnoxious calls to action? [1]: https://i.imgur.com/eIhwtjo.png

A similar discussion can be found here [1] as well which also got to front page as well. The content is strikingly similar : detail OCR introduction and goes to API introduction.

[1]: https://news.ycombinator.com/item?id=21626025

Re: Automating receipt processing with deep learning

#19

Post by amelius »

TrackerFF wrote: post #16
amelius wrote: post #10

Earlier quoted context omitted.

> Need to digitize documents, receipts or invoices but too lazy to code? Head over to Nanonets and build OCR models for free! I think that once you go down the path of OCR'ing your own invoices, you better start a company around it.

Gonna be honest, I think it's a dead-end. The solution IMO will come from the supply side; Easier to get receipts digitally from the store / vendor / seller, than to spend all this effort on converting physical receipts to digital. Sure - there's good retro usage, I'd rather push for sellers to offer you digital receipts. In fact, I already get that from my local supermarket. My bank card is registered to their app,…

> Receipt management was one of those entrepreneur ideas that would always pop up.

Not surprising as it's one of the first things one runs into after starting a company.

Re: Automating receipt processing with deep learning

#20

Post by mywacaday »

tastyminerals wrote: post #7

It's a nice overview article for anyone interested in the topic of IE from financial documents. However, for industrial level solutions Tesseract does not cut it. Abbyy is the best OCR engine on the market currently. Receipt IE works just fine with rules supported by a small BiLSTM as fallback just because receipts do not contain a lot of text. With invoices this approach is suboptimal. On a general note no DL approa…

I had abbyy a long time ago and had some uses for it so i went to the website to check out the cost and was met with this monstrosity "Protect your shopping cart downloads with Download Insurance Service. For only £11.00 you will be able to download your files for 24 months, in case you need to reinstall the products. ABBYY Screenshot Reader". Ripoff!!, I've never seen anything like it elsewhere, has anybody else?
Post reply on HN