Artificial Neural Networks Android-Based Interface Facial Recognition Systems

Abstract

Face recognition system is a crucial issue these days. This research builds an Android-based facial recognition system in real time using the Gabor filter and artificial neural network (ANN) methods. The system can be implemented properly. The test results show that for testing in scenario 1, the largest accuracy is 90% in hidden layer 4 and 5. The smallest computation time is 0.46872 seconds for layer 2 and the biggest time is 0.63778 seconds for hidden layer 5. While the test results for scenario 2 shows the lowest accuracy is the trainrp training function for 76%, while the highest accuracy of 94% is in the traincgp training function.