Fork me on GitHub
#off-topic
<
2016-10-21
>
roelofw12:10:50

Is it possible to use machine learning so I can scan invoices and use it in my to build financial app ?

sveri12:10:30

@roelofw I just read about a major certificate breach at comodo https://www.heise.de/security/meldung/Zertifikats-Klau-Fatale-Sehschwaeche-bei-Comodo-3354229.html (article in german). The gist was, they used OCR for scanning forms which, well, did not work to well. So for invoice stuff and friends I am sure you can do that, but I myself would not trust it for now.

roelofw12:10:46

oke, then maybe my idea is not so well. I was thinking about it because I know some major cloud accounting apps can do it

sveri12:10:39

It surely can be done, but then again, would you trust it in the end? Would you bet on it being right 100% of the time?

roelofw12:10:33

I do not think so, that why the most have a screen where the user can change the data which is not well

roelofw12:10:14

I never heard of a ocr or machine learning which is 100% right

sveri12:10:20

Yea, me neither. Although, maybe it would be enough to be better than the typical user entering data. Hm, not sure what their error quote is

roelofw12:10:33

I also not

roelofw12:10:50

I was just wondering if I could do the same

sveri12:10:04

I think it would be fun to try that

roelofw12:10:08

but maybe they use other programming languages

roelofw12:10:10

I know but I have to have a starting point and I cannot find good info on ocr and/or machine learning on this subject

roelofw12:10:43

I rather use something like haskell or clojure because I love FP languages

sveri12:10:12

I would bet there are libraries for Java enable that you could wrap

roelofw12:10:43

I have not find them 😞

val_waeselynck13:10:14

@roelofw maybe if you find an implementation in C / C++ you can wrap it with Pixie https://github.com/pixie-lang/pixie

roelofw13:10:21

@val_waeselynck I found tesseract but I have then find out how I can make it work with a invoice

roelofw13:10:35

The example is only talking about a webcam

val_waeselynck13:10:09

@roelofw I have no experience in this domain

val_waeselynck13:10:52

But personally, if I found a well-supported Python lib for it, I'd use that, the programming language is the least of your problems for this stuff

sveri18:10:24

hm, is github down for anyone else?

sveri18:10:00

thanks, seems like netflix is down too

sveri18:10:11

how should that work, if I cannot code and cannot watch something

sveri18:10:14

ahhh, what to do?

bradleyc18:10:48

If you need access to GitHub, add these to your hosts file:
192.30.253.113 
151.101.4.133 

sveri19:10:27

Unbelievable how many links lead to github

akiroz19:10:18

Try different DNS servers, the OpenNIC network seems to be working with github.