Advances in Decision Forests and Ferns with Applications in Deep Representation Learning for Computer Vision

In recent years, deep learning has largely embedded itself within computer vision as an extremely useful tool for accomplishing a range of vision-based tasks such as image classification, object detection and semantic segmentation. With this surge in popularity in deep learning, more traditional computer vision methods such as decision forests have somewhat fallen out of favour in the community despite their benefits. This thesis hopes to bridge this gap; it focuses on advancing the ensemble methods of decision forests and ferns and incorporating them within deep learning frameworks for computer vision applications.