stationary target shooting algorithm [closed]

asked 2012-08-08 17:20:54 -0500

psychobyte gravatar image

updated 2018-01-07 17:34:59 -0500

Hi,

I'm totally new to opencv and my c++ isn't too great. That said, I'm trying to use the OpenCV python bindings(for now) to create a real-time application that will track when a resetting shooting target is hit and keep score. I'm hoping to get some suggestions on how to develop the algorithm.

so basically, the targets are circles or perhaps "Lollipop-shaped"(i'm sure squares will be used as well). When they are shot they will either drop out of view or perhaps switch sides on the platform.

For the first phase, I'd like to be able to identify the targets(target 1, target 2, target 3, etc) by position and give a score when it is hit (goes down). The targets can vary in size based on how far a way they are (25-50 yards). The targets can reset during a shooting session so "object permanence" would be nice as well.

I'm practising with object detection through tutorials (python) right now using still images and recorded video. Can anyone offer some suggestion how to tackle this problem?

Thanks!

edit retag flag offensive reopen merge delete

Closed for the following reason the question is answered, right answer was accepted by sturkmen
close date 2017-09-04 16:08:54.151990

Comments

Could you please attach some images? It is possible that your targets can be segmented by color, you can search this site for questions about color blob detection, i.e. http://www.answers.opencv.org/question/265.

Kirill Kornyakov gravatar imageKirill Kornyakov ( 2012-08-10 01:40:42 -0500 )edit

Hi,

this is one example http://www.centerfireguns.com/do-all-traps-resetting-target-ap22.html

i suppose segmenting by color is not practical because after getting shot they begin to lose their coloring. Although, i suppose it may not be uncommon to re-spray them to refresh their coloring.

I'm thinking that I could order them by by centroid positions. (left to right, 1,2,3, etc)

psychobyte gravatar imagepsychobyte ( 2012-08-10 13:23:03 -0500 )edit

It is an interesting problem but you may have problems trying to find a solution for every kind of target. Maybe you should focus on one kind of target and add complexity later.

Rui Marques gravatar imageRui Marques ( 2012-08-13 08:35:16 -0500 )edit