Research Topics
[BA] Evaluation of Face Recognition Cloud Services on Mobile Devices for Human Identification
Motivation: Modern software technology offers many implementation opportunities for Ma- chine Learning (ML) in terms of automated decision making, speech and text synthesis or, the main interest of this thesis, face recognition. Biometric analysis for personal identification have been developed quickly in the past decade. Fingerprints are already widely used in many systems and software applications. It can be as- sumed that in the nearest future face recognition will be also widely adopted for the purpose of human identification. In this case, a mobile application can serve as a portable and effec- tive solution. Although independent implementations of face recognition algorithms directly on mobile devices are possible, it can be more time and cost efficient to use cloud services, such as Amazon AWS, Microsoft Azure etc., which provide a wide range ML technologies and up-to-date APIs for easy integration in client applications. Context: A possible context of such an application is the management of visitor access to buildings. It is important for some organizations to keep track of its visitors and collect their data upon every visit. The conventional solution is to fill out a form to get a visitor access. To simplify this task, a face recognition system on a mobile device can be utilized to collect person’s data on his first visit, re-use it to quickly identify this person and keep track in a digital log book upon every next visit. Goal: The goal of this thesis is to carry out research on the current state of face recognition and its integration within mobile devices. To accomplish this goal, it is necessary to gain a deeper understanding of face recognition from scientific research, to find out and compare existing cloud ML services as well as to create and test an implementation to integrate these services using mobile Android devices.
Betreuer: Karsten Wendt