Design of Face Recognition Systems in Complex Environments Using CNN
Design of Face Recognition Systems in Complex Environments Using CNN
Diana Jamshidian1
1) Bachelor of Science in Computer Software Engineering, Department of Computer Engineering, Abrar Institute of Higher Education, Tehran, Iran
IOI Code
:
AZEK-AACDDZ
Publication :
4th International Conference on New Research & Achievements in Science, Engineering & Technologies(setbconf.com)
Abstract :
Face recognition in unconstrained environments poses significant challenges due to variations in illumination, occlusion, background clutter, and facial expressions. This paper presents a deep learning-based approach for designing a robust face recognition system using Convolutional Neural Networks (CNNs). The proposed method leverages the feature extraction power of CNNs to learn discriminative facial representations that remain effective under complex and unpredictable conditions. We conduct extensive simulations on standard benchmark datasets and artificially generated complex environments to evaluate system performance. Experimental results indicate that our CNN-based architecture achieves high recognition accuracy and strong resilience compared to traditional face recognition techniques, demonstrating its potential for deployment in real-world applications.
Keywords : face recognition convolutional neural networks multi scale feature fusion spatial attention mechanism data augmentation complex environments