HuskyLens looks great and similar to an idea I already had…so someone has done the work for me!

However I want to know, will Husky lens work with my use case? Here is what I would like it to do:

I would point camera at vehicle dash board, in the area where gear selection is displayed (i.e. P R N D). I would teach it what P R N and D look like when the gear is selected.

I think this would work fine for instance where gear is shown by single graphic and the graphic is changed entirely when gears are changed (i.e. P changed to N or N changed to D).

But would this work for the case where:

P R N D are always displayed on the dash board but selected gear is highlighted instead in order to show selection? (i.e. When D is selected, D is displayed in Red or is made Bold, or there is “selection box” round P). For example of this, Tesla Model 3 (https://i.redd.it/ha74ex068co01.jpg) shows P R N D all times, but selected gear is just made bold. So I would like “D” to be identified, but not the other letters.

VW ID3 might be tricky (https://car-images.bauersecure.com/wp-images/12852/id3_035.jpg) because it is white letters on black background all the time. But selected gear is highlight in orange. Unless “P” is selected, then it is parking brake icon highligh in red.

I feel like the “Tag” recognition mode might be best suited to this application, even though they are not tags. But usually gear display will be simple, clear, high contrast