Computer Vision with OpenCV 3 and Qt5

Speed

How fast your algorithm performs is especially important, in case you are developing a real-time application where the FPS (frames per second, or framerate) value must be as high as possible. So, same as the accuracy, you need to be careful with clarifying this requirement, too. If you match two images and show some matching results to the user, even a delay of half a second (500 ms) may still be acceptable, but, when working with high FPS values, half a second of delay for each frame is extremely high.

You can use the TickMeter class or the getTickFrequency and getTickCount functions in OpenCV to measure the execution time of a computer vision process (or any process for that matter). First, let's see how the older method works: