Another technique we'll be using is Pixel Intensity Comparison-based Object detection (PICO), originally developed by Markus, Frljak, et al. in 2014. It uses the same broad principles as the Viola-Jones method, in that there is a cascade classifier. It differs in two ways. First, a sliding window is not used. This is due to the latter differences. Second, the classifiers of the cascade classifier are different from that of Viola-Jones. In Viola-Jones, a method of applying filters repeatedly and then summing the result is used as a classifier. By contrast, in PICO, decision trees are used.
A decision tree is a tree where each node is a feature, and the branching of the feature is defined by a threshold. In the case of PICO, the decision tree applies for each pixel in the photo. For each pixel considered, the intensity is compared against the intensity of another pixel...
