The focus of this report is to investigate the relation of Face Recognition (FR) and Machine Learning (ML), since technology is continuously progressing, making it easier to incorporate computers to execute this seemingly trivial, to humans, task. A FR system can more efficiently incorporate ML as a learning framework, in order for the system to perform the task it is assigned to. Nevertheless, such systems are still not used as much as they potentially could, which lead to the motivation of this project: how ML has been applied to FR systems, what issues are these systems facing and how can they be dealt with. An introduction to the actual terms of FR and ML will be given, from the historical and theoretical point of view. The chosen algorithms that are thought to be the most prominent and worth investigating will follow, and then these algorithms, namely Eigenfaces, Artificial Neural Networks, Pose Estimation, Local Binary Patterns Histogram, and Infrared, will be briefly described and explained. A comparison of their performance and applicability will follow, finishing with the discussion on how can they be used and combined optimally. We suggest a combination of the methods available for a higher accuracy, despite the additional computing power needed.
|Uddannelser||Basis - International Naturvidenskabelig Bacheloruddannelse, (Bachelor uddannelse) Bachelor|
|Udgivelsesdato||19 dec. 2017|
|Vejledere||Ole Torp Lassen|