University of Basra's Master's thesis on recognizing masked faces
A master's thesis at the Faculty of Computer Science and Information Technology at the University of Basra researched the recognition of masked faces using a Siamese neural network
The thesis, submitted by master's student Ali Haitham Abdulamir, aimed to improve the performance of masked face recognition systems using a Siamese neural network, by integrating it with pre-trained models namely: Xception, MobileNet, EfficientNet, and ResNet-50 for deep feature extraction and image comparison for identification.
The thesis used data balancing to reduce bias and improve generalization to previously unseen data, enhancing the performance of the model in real-world environments. The results showed high effectiveness when combining Siamese network with these models.
Department of Media and Government Communication
