Facial key features detection algorithm implementation

The goal of this thesis is to develop an algorithm for detecting facial key features on grayscale images. The task is complex and has the potential to be implemented in many fields such as medicine, security, automotive industry and gaming.;The code was written in Octave. This language is optimal for prototyping. It has a simple syntax and powerful statistical libraries.;This algorithm models a neural network for key feature patch detection using a sliding window and a linear regression for identifying the rest of the features. Obtained results generate a root square mean error around 4.28 and are satisfactory compared to the best results that are under 2.00. The model could be refined using a deep convolution neural network.