A MULTI RESOLUTION CONVOLUTION NEURALNETWORK BASED FACE RECOGNITION ANALYSIS

Face recognized image processing and biometric systems is one of the most efficient and relevant applications. This paper explores the methods of facial reconnaissance, the algorithms proposed by several researchers in the field of image processing and design reconnaissance using artificial neural networks (ANN). In this article, we will also explore how ANN is used for the face recognition system and whether it is better than other approaches. Two steps are used to develop the facial recognition system. The first stage is to take or remove the facial features and the second step is to identify the pattern. Deep learning, especially the CNN, have made commendable progress in the field of FR technology in recent times. This paper looks at the performance of the pre-trained CNN with the SVM classification and at transfer learning results using the AlexNet model to perform classification. The paper is available in French only. The study examines CNN architecture, which in recent years , specifically AlexNet and ResNet-50, has recorded the best results at the Large Scale Visual Recognition Challenge (ILSVRC) in the ImageNet. Recognition accuracy has been used as a determinant for evaluating output optimization of the CNN algorithm