Is the use of AI possible in CAPI Text Data Collection from Image?

[vgonzalez]

The use of AI is being considered for data collection from images to complete data and reduce errors when collecting data in the field, specifically in unstructured or open text variables such as First Name, Last Name, Address, Date of Birth from an identification document, for example, from a driver's license.

ID_Sample.png

First Name: IMA
Last Name: CARDHOLDER
Date of Birth: 08/31/1977
Address: 2570 24TH STREET ANYTOWN, CA 95818

[Gregory Martin]

We haven't looked into adding anything like this to CSPro, but it could be possible to add a wrapper around something like Tesseract so that text could be extracted from images:

https://github.com/tesseract-ocr/tesseract

What is your use case?

[htuser]

@vgonzalez
Ah Ah... I worked on the same issues, nearly four(4) years ago. Please take a look at my post here: viewtopic.php?p=14388
You can also read: viewtopic.php?p=18675

More recently on the web from the JS ecosystem:
a) A recent JS Compatible use case can be read here: https://transloadit.com/devtips/integra ... seract-js/
b) Working example : https://github.com/simonw/tools/blob/main/ocr.html
c) Tesseract alternative: https://github.com/scribeocr/scribe.js/tree/master and comparisons: https://github.com/scribeocr/scribe.js/ ... sseract.md

However Worker Thread action invoker support by the CSPro Developer Team is very important for such operations in CSPro powered applications.

Josh is implementing a very innovative feature that will boost CSProusers productivity using similar and more advanced ML models: https://github.com/jhandley/survaize


With the complete JS-CSPro logic bidirectional API of the future CSPro 8.1, such things will be more portable to the CSPro ecosystem.
I'm waiting the alpha or beta version version for some testing.
Best

[vgonzalez]

Thank you both very much, I'll take a look at the previous posts and see how I can start a project from zero to see how it goes.