In recent years decision forests have established themselves as one of the most promising techniques in machine learning, computer vision and medical image analysis. This review is directed at engineers and PhD students who wish to learn the basics of decision forests, as well as more senior researchers who wish to push the state-of-the-art in automated image understanding. The authors present a unified, efficient model of random decision forests, which can be used in a number of applications such as scene recognition from photographs, object recognition in images, automatic diagsis from radiological scans and document analysis. Such applications have traditionally been addressed by different, supervised or unsupervised machine learning techniques. In contrast, this survey casts diverse tasks such as regression, classification, and semi-supervised learning as instances of the same general decision forest model. The flexibility of the forest framework further extends to tasks such as density estimation, manifold learning, and semi-supervised learning. The unified forest framework allows for the opportunity to implement and optimize the underlying algorithm only once, and then easily adapt it to individual applications with relatively small changes. The theoretical basis and numerous explanatory examples presented in this survey serve as a solid platform upon which to build exciting future research.