Face detection in a real-time videostream on Raspberry Pi

This paper is focused on the topic of the system of face detection in a real-time videostream on a low-power device. It describes the implementation and differences of two selected approaches, specifically the use of CNN or Haar classification. It explains the importance of the data preparation, especially the process of modifying the input images. Also, it deals with the problems and limitations arising mainly due to the choice of using a low-power Raspberry Pi device.