Newbie questions

Hi all.
I have been recently working on a project in which our goal is to identify single uppercase letters printed out on a wall, when I stumbled upon this site, for the first time.
I have never worked with OpenMV before, so I would like to ask:

  • Do you think it would be possible for the camera to do the procedure in realtime by itself (varying factors being position, lightning and hue)? If yes, which functions did you have in mind?
  • If not, would it be possible to first take a screenshot when the camera sees something black, and then transmit that to something such a raspberrypi for it to use as an input for a sophisticated classification program such as Google’s TensorFlow?

Btw, even if all answers are negative, you’re already my favourite kickstarter project of all time :wink:

Hi, yes this can be done with template matching. Basically, you take a picture of the letter on the wall and then compare that picture to the view of what the camera sees in real time. You’ll be able to do this in real time.

That said, template matching works to find just the template… So, it’s not rotation or scale invariant. It will find the template in the image but only the best possible location of the template.

Huge thanks for the answer!
The letters will always be displayed upwards, and the scale almost the same… so it should work! :smiley:

Gonna grab my cam soon.