Over the last decades, the processing of video and images in real time seemed impossible due the big amount of data, but now with the new generations of processors and cloud services it is possible and feasible.In this document, we will explore the usages of deep learning and machines learning for image categorization based on the genre and the mood. One of the goals of this projects is basically know how the facial recognition techniques works. The main idea is to modify the input parameters and do some improvementson an existing library and also recognise the strong and weak points, and if it is possible to increase the performance and the rate of success using mainly image pre-processing techniques.For that I will use anexisting open source library, in order to be able to make those improvements and check how those changes affects the resulting image.Furthermore, I want to learn how the recognition models evolve, in order to see how the chosen model impacts in the finalresult, and how works the autonomous learning process.In order to have a reference, I will use a professional face recognition tool to check how our library behaves and how accurate it is.