(Archived) suggestion: simple OCR correction

[soundsgoodtome 147]

the OCR feature's great, but can be somewhat overzealous. that's ok -- it's a computer -- and i'd rather deal with false positives than the opposite.

however, when we discover incorrect recognitions, we need a basic way to tell EN "that is not the word you thought it was." on the mac client, preferably we'd control-click to bring up a contextual menu then have the option to select either "permanently remove" or "correct detection."

i'm attaching one that came up today. i searched for notes containing the word "timer," and EN thinks it sees the word "timer" in the pattern of the berber rug. :-) again, that's ok: it's a computer. but we should have a method for correcting EN, so it never comes up again.

if my image had in fact contained a word, but not the word "timer," i'd have wanted to tell EN what the correct word is. but in this case -- since there's no word at all -- i'd like to be able to tell EN "that's not a word (i.e., ignore this from now on)."

i'm surprised and frustrated that this functionality doesn't yet exist. it's not frustrating that a computer can't "see" as well as humans; it's frustrating that we humans haven't been given a way to correct the computer. :-/

[gbarry 2,657]

I'll have to take a deeper dive into our OCR capabilities and OCR tech in general to see if this is possible/implementable, but thank you for the suggestion.

Personally, I see patterns all the time in rugs, so I sympathize with the computer here.

[jefito 5,586]

Personally, I see patterns all the time in rugs, so I sympathize with the computer here.

I used to as well. Then I graduated from college.