ETD system

Electronic theses and dissertations repository


Tesi etd-05132015-110009

Thesis type
Tesi di laurea magistrale
Design and prototyping of a face recognition system on smart camera networks
Corso di studi
relatore Prof. Amato, Giuseppe
correlatore Prof. Falchi, Fabrizio
correlatore Prof. Marcelloni, Francesco
Parole chiave
  • PCA
  • LBPH
  • face recognition
  • OpenCV
  • Smart camera networks
  • face detection
Data inizio appello
Riassunto analitico
The aim of this work is to design and develop a face recognition system running on smart camera<br>networks. In many systems, these are passively used to send video to a recording server. The<br>processing of the acquired data is mainly executed on remote and more powerful computers (or<br>clusters of computers). In this thesis a distributed architecture was developed where computer<br>vision algorithms are executed on smart cameras, which can exchange information to improve<br>resource balance. A smart camera network has been defined specifying the roles that client<br>nodes and a server have, and how nodes cooperate and communicate among them and with the<br>server. Smart cameras, initially look for changes in the environment. When motion is detected,<br>they perform face detection. Once a face is found, the camera itself processes it and tries to<br>asses whom it belongs to, using a local cache of recognizers. This cache stores a portion of the<br>whole information present on server side, and can be used to perform recognition tasks on the<br>smart cameras. If a node is not able to identify a face it sends a query to the server. Finally,<br>if the person’s id can be determined, either by the server or the client itself, the occurrence of<br>the correspondent recognizer is notified to the nearest nodes. Human faces that were not recognized,<br>are stored on the remote server and can be manually annotated. Clustering algorithms<br>have been tested in order to automatically group faces belonging to unknown people on server<br>side so they made the manual annotation easier. Extensive experiments have been performed<br>on a freely available dataset to both assess the recognition performance and the benefits of<br>using collaboration among cameras. Raspberry PI devices were used as camera network nodes.<br>Various tests were performed in order to verify the efficiency of the face recognition approach<br>on such devices.