This chapter explains how to use the different options included in the OpenCV object detection module. With the sample code included, it is possible to use Cascade and Latent SVM detectors as well as create custom cascade detectors for a specific object detection application. Additionally, the new Scene Text Detector included in OpenCV 3 is explained in the chapter.
Object detection deals with the process of locating instances of a certain class of real-world objects, such as faces, cars, pedestrians, and buildings in images or videos. Detection algorithms typically start by extracting features from two sets of images. One of these sets contains images from the desired object and the other one contains background images where the searched object is not present. Then, the detector is trained based on these features to recognize future instances of the object class.