Face recognition : implementation of face recognition on AMIGO

In this (traineeship)report two possible methods of face recognition were presented. The first method describes how to detect and recognize faces by using the SURF algorithm. This algorithm finally was not used for recognizing faces, with the reason that the Eigenface algorithm was an already tested and proven method for facial recognition, which did not hold for the SURF method. This does not mean that SURF would not provide good results for face recognition. From the test data it was clear that SURF had some good results, but as mentioned before these results did not always have the same quality. The second method was the Eigenface method. Because this method proved to be capable of recognizing faces, it was implemented on AMIGO. The algorithm was used during the RoboCup@home challenges in Germany. It was capable of acquiring training faces, which was the first important step. After this acquiring step, due to a combination of an error in the world model and face detector, the algorithm ended up in an endless loop. Further results of the actual implantation on AMIGO are therefore still unknown.