Need help looking for a set of symbols in a video.
Hi, I am currently working on a school project where the goal is to create a "robot" that would be able to orientate itself using a set of symbols, those can be anything. What I've tried so far is template matching but it was without success.
There were two main problems
- the symbol just wasn't recognized from in the video
- I have no idea how to create conditions based on the results of matchTemplate(). What I've tried was:
...
res= cv2.matchTemplate(gray, template, cv2.TM_CCOEFF_NORMED)
...
if(res.any()): "some result"
or
...
res= cv2.matchTemplate(gray, template, cv2.TM_CCOEFF_NORMED)
...
if(res.all()): "some result"
but neither worked. At this point I'm looking for any suggestions on how to accomplish this. The last resort solution I would go with would be looking for Russian license plates using haar cascades and reading the numbers inside them, then performing some action based on the numbers.
idk, how good your teacher are with computer-vision. but template matching only can apply, if there's no rotation/perspective/scaling.
(probably the wrong hammer, if you have a robot, which gets it s input from ", running around" (involving perspective and all)
The thing is that they don't teach neither python nor any kind of cv in our school, I just chose it because it seemed cool (Yes, I am an idiot). So template is not an option here, guess I'll have to go with the Russian plate option because I don't have the time or skills to make any kind of deep learning algorithm or something like that
@Nillas, -- i totally sympathize with you. but again -- your question is missing the essential context, like: what are you trying to achieve here ? (that makes it an "XY" problem, look it up, please)
(you mention number plates, --what's * your* state of of research on that ?)
the goal is to create a "robot" that would be able to orientate itself using a set of symbols, those can be anything.This is your statement. So, I would go with an Aruco marker. You will be able to detect accurately and robustly the marker and you can use also different tag ids.
@Eduardo that might be exactly what I was looking for, thank you. It even comes with some kind of a mapping function if I'm not mistaken so that's super great