Performance Evaluation for Infrared Face Recognition using Convolutional Neural Network
Muhammad Eka Setio Aji, Annisa Syakhira, Supriyanto Praptodiyono, Rocky Alfanz · 2022
Infrared image has a potential to be integrated with face recognition system because the ability cannot be affected by the illumination. Thermal camera can capture image in low light condition, but a low information and dept texture quality still a weakness in a problem of infrared face recognition. Convolution Neural Network can extract more depth the features with the convolution layers. Convolution layers are needed to extract the feature from the infrared image with a low information. The combination of Convolution Neural Network with Haar Cascade can improve the result of evaluation rate. In this work, we present a comparison from several transfer learning model with ERA6-Net as our proposed model. ERA6-Net with a simply layer show superiority in the evaluation model in several parameters with the highest score in accuracy up to 95%.