EFFICIENT FACE DETECTION USING VIOLA-JONES ANDNEURAL NETWORKS: A COMPARATIVE STUDY

Abstract

 Face detection technology underpins most of today’s face recognition systems and is
valuable in many industries, including security, healthcare, retail, and entertainment. This study aims to
fuse classical computer vision methods, like the Viola-Jones algorithm, with contemporary techniques in
deep learning, for instance, Feed Forward Neural Networks (FFNN), to improve face detection systems in
accuracy, efficiency, and reliability. The proposed system uses the Viola-Jones algorithm for preliminary
face detection and an FFNN for feature extraction and classification achieving 98.5% accuracy on various
datasets. The system works well under a variety of conditions such as lighting, angle, and occlusions, and
has real-time performance with frame rates between 15-20 FPS. Results confirm that the system is more
accurate than applying the Viola-Jones method alone and has the same accuracy as CNN-based models
while needing less computational resources. This approach is useful and effective for practical applications
like security surveillance, biometric identification, or human-computer interaction because they are more
rapid and easier to deploy.