Behind scenes involves in
1. OCR - to detect characters
2. use location bindings of the words as input to an AI model which spits out table cell ids that contribute to form a table structure
- Decent UI
- conveyed the purpose from the site
However, a trail/demo page to test out an image from UI would be much helpful than expecting people to try out with the code.
I agree, this was inline with the next improvements. As this looks like a priority, you should be seeing this in a day.