Development and analysis of embedded face recognition system using Raspberry Pi

• Main Author: Falah Hassan, Alwan
• Format: Thesis
• Language: English
• Published: Universiti Malaysia Perlis (UniMAP) 2019
• Subjects: Human face recognition; Raspberry Pi (Computer); Embedded system; Face recognition system; Face recognition system > Design and construction
• Online Access: http://dspace.unimap.edu.my:80/xmlui/handle/123456789/61835

Human Face is the most visible part which can be used to recognize persons. There are many available systems for face recognition in the market, but they are bulky and expensive. The implementation of face recognition techniques in an embedded system is a very important aspect. This project involves design of a real-time, portable, low embedded cost face recognition system. Implementation and analysis of face recognition techniques on an embedded system, the development phase consists of Single Board Computer (SBC, Raspberry Pi (Model A) as process unite, and GNU/Linux based Embedded Raspbian Operating system is used as application development platform. This project focuses to apply the face recognition algorithm that is suitable with Raspberry Pi (Model A) The proposed system is implemented using ARM11 processor and inefficient memory on Raspberry Pi (Model A) board, to get an acceptable performance of the system, the images are captured at resolution (320×240), the system needs ≈ 2.1 sec to process the captured images, The performance of the embedded system is done by evaluating detection time and recognition time (is 1.75 sec, between 0.29 sec to 0.74 sec) respectively, together with CPU utilization and RAM utilization (33%, 17.75%) for detection and (36.5%, 22%) for recognition. Results obtained shows that the overall performance on the embedded system can be increased when motion detection techniques is applied.