Description of Invention
To improve the precision and robustness of the system, the 2.5D face recognition system makes use of the depth characteristic found in the 2.5D data (depth image), which provides additional information. In this research, we proposed a fusion-based 2.5D face recognition using EfficientNet-B4 with various optimizers. The performance of the 2.5D face recognition can be enhanced by combining the feature fusion method with deep learning approaches. This is due to the feature fusion techniques, which fuse the features extracted from the depth images containing discriminative data that can increase the accuracy rate. The main goal of this project is to evaluate and compare the performance of these feature fusion method to find out which fusion method performs the best.