Detecting animated VICTORY! or DEFEAT text from match end screen


I really need help with the current project I am working before i go crazy. Firstly, my background in programming is about enough to get me in trouble, I have also never used openCV, so to say i am struggling would be an understatement.

I have been trying to determine the best method in capturing the word VICTORY! or DEFEAT from the “match end” screen in my game. I had been using sample images for testing which I have included 2 links of the images I have been using for reference (I am a new user and could not upload more than 1 image or include more than 2 links, sorry).


I have been using chatGPT to generate code to test different openCV and tesseract methods, though results are questionable and i feel my “assistant” is not utilizing optimal methods.

Can someone please give some advice on the best way of managing the text in the supplied images. Sample code would be amazing though is not expected, though ultimately I am just hoping to find some guidance amongst this darkness, to know what direction I should be heading in.

Any help is greatly appreciated and i will eternally love you! :smiley:

template matching should be good enough.

opencv’s template matching is not “multiscale”. template must be the exact same scale as its instance in the “haystack”. the more it is not, the more the score will differ from a perfect match.

you can reduce the haystack to the exact size of the template because you know where the template will be, just not whether it’s there.

template matching score will say whether template is there.