I'm currently planning out an application that takes a video input, searches for a pre-defined event (an existing image) in said video and then triggers actions upon locating a defined event. The actions I am currently looking at are to take the frame where the event occurred, save it as an image and call an API.
I have already been looking at OpenCV's template matching as a possible solution, but so far I have only found resources discussing real-time searching. As my video input could be 1-3 hours long, Ideally this is not done in real-time.
Is it possible to use OpenCV's template matching for this purpose or are there other better-suited alternatives?