Low cost and processor efficient facial recognition system

摘要

The objective of this research project is to develop a system with a webcam and efficiently detect faces of persons preciously loaded in the database. This project aims towards the use of facial recognition using mundane processors and hence the system is optimized and reviews the grand challenges that are faced for developing such a system and a literary survey that has provided a crucial insight in the developing of the system. Viola Jones Algorithm has been utilized to detect a face within a picture and ECOC for training to fit multi class models into categories or code-words. With the algorithm and the code as suggested in this paper, results are projected towards various lighting conditions of front profile of faces. Under normal lighting conditions a very high detection rate of has been obtained. Under partial exposure a lower detection rate has been obtained and at overexposed conditions detection rate similar to normal lighting conditions has been obtained. The rejection rate is however, cent percent in all three conditions mentioned. There is ample scope for implementing recognition techniques.