Abstract
ANNs have the ability to learn from data and recognize unstructured patterns. In face detection, ANNs can be used to recognize faces from images or videos. However, selecting the right ANN method is crucial for achieving optimal face detection performance. The results of this study are the highest classification accuracy value of the backpropagation artificial neural network system in recognizing student facial images with a comparison of the Learning Vector Quantization method is 96.25% (detection error = 0) with a classification architecture of the test data comparison - training data 70: 30, 20,000 iterations, calculation error tolerance 0.00001 and the number of hidden layer neurons = 5 while the lowest classification accuracy value of the backpropagation artificial neural network system in recognizing student facial images based on LVQ features is 85.42% (detection error = 34) with a classification architecture of the test data comparison - training data 70: 30, 10,000 iterations, calculation error tolerance 0.01 and the number of hidden layer neurons = 15
