Recognising features from faces is a fundamental requirement for any system that hopes to analyse faces. In general, facial feature point detection algorithms attempt to find a sparse set of points on a face that correspond to definite features, such as the tip of the nose. Given the expressive nature of our faces and the large variety of poses present in the average photograph, facial feature point detection remains a challenging task.
In this seminar I aim to provide a general overview of the spectrum of techniques that have been proposed for facial feature point detection. In particular, I will focus on state-of-the-art techniques that have been implemented by members of the Intelligent Behaviour Understanding Group (IBUG) within a new Python package called Menpo. Menpo aims to provide a simple yet powerful playground for exploring image data and is useful for anyone who manipulates images on a daily basis. I will focus on simple examples of building and utilizing different models for facial feature point detection. However, Menpo is not a face specific framework and thus is useful for any kind of object that requires feature detection. Finally, I will talk about the current state-of-the-art for mobile devices and the future of facial feature detection.