Computer Vision with OpenCV 3 and Qt5

Accuracy

First of all, accuracy is quite misleading, because, as soon as we see some accuracy drops, we usually tend to move away from an algorithm, but the correct way is to clarify the accuracy requirement for your use case first. Check out the data sheets of the computer vision-based machines made by very well-known companies, and immediately you'll notice things like more than 95 %, and so on. This doesn't mean that the machine is imperfect. Quite the opposite—it means the accuracy of the machine is well-defined and the user can expect a certain amount of accuracy, and, at the same time, they can live with a certain low amount of error. All being said, it is always good and recommended to aim for a 100 percent accuracy.

There is no better ...