Hacker News new | ask | show | jobs
by MathYouF 1753 days ago
I feel like there's a deep learning (possibly computer vision) route to this that would be extremely hard to create but would be able to arbitrage 90% of the difficulty with updating for new UIs or handling edge cases. Had you ever pursued that route or seen things that tried?
1 comments

I really don't understand your usage of the word arbitrage here. Also computer vision is related to images specifically - how is that even relevant here?
This guy can speak for himself but I've read the comment a few times and I would surmise there's some ESL word-soup happening here. I am going to hazard what he's getting at is using CV to import data from one UI possibly into another. I think to get around the lack of a lack of API support. Similar to the effect a company might use OCR software to migrate data from paper forms a CRM, or something? I'm still wondering why I honed in on this one comment, but hey you felt intrigued enough to leave a comment too, so there must be something to it.
What does ESL mean? I can't seem to infer it
English as a Second Language
English as a second language I presume
English as a Second Language