Ask HN: Is Deep Learning reasonable for a 1 man team?

I'll explain the problem:

>I need to collect store prices(and item name), but this requires taking a photo

>This needs to be typed and lined up to internet data that has some flaws.

-The names are incorrect (fuzzy text is correct only 90% of the time)

-Sometimes there are combinations/deals that require 4-8x as many extra combinations. I believe this needs to be figured out by a person.

Everything else is automated after that.

Currently using python, but I have 10 years of programming experience, and working on a few other unrelated projects in other languages.

This idea to use Deep Learning seemed like such a weird buzzword that I blew it off. Looking into it, it might???? be reasonable?

Is this worth my time? or is "Keras" more effort than its worth?