Any computer vision problem can be solved in different ways. Each way has its pros and cons and relative measures of success, depending on the data, resources, or goals. Working with OpenCV, a computer vision engineer has many algorithmic options on hand to solve a given task. Making the right choice in an informed way is extremely important since it can have a tremendous impact on the success of the entire solution, and prevent you from being boxed into a rigid implementation. This chapter will discuss some methods to follow when considering options in OpenCV. We will discuss the areas in computer vision that OpenCV covers, ways to select between competing algorithms if more than one exists, how to measure the success of an algorithm, and finally how to measure success in a robust way with a pipeline.
The following topics will be...
