Find postion of irregular item using 3 cameras around a dartboard

Python
1

Hi all,

I’m working on a personal project where I want to have an automated dart counter. So the app should film the dartboard, recognize a new dart arrow, try and find it’s position which can be translated into a score. This is my first encounter with computer vision so it is a bit of a challenge.

What I’ve done is put pins around the dartboard to have fixed points to rectify the image so every camera angle gives the same output (clockwise: 20 > 6 > 3 > 11)

I think theoretically this must mean that the common point of all 3 different images is where the dart enters the board, thus giving me the location. I however have a lot of trouble ‘finding’ the dart programatically as it’s an irregular shape. Also it does not help that a dart may obstruct the lighting, so that gets recognized as a change as well + the dartboard may move ever so slightly on impact making it difficult to find only the edges of the dart.

Anyone with ideas on how to approach this problem? Thanks!

2

That’s a fun project

I would recommend something simple to start with.
Your dartboard is a plane, so plane to plane transformation is an homography. If you know the size of the dartboard a priori, you have everything you need. One camera would also be enough.
Then detecting the darts could be translated as comparing before/after? Obviously you can also train a dart object detector.

I wish you the best on your learning journey
G.