The presented study proposes an enhancement to the Eigenface algorithm used in face recognition tasks, specifically an attendance management system. Traditional implementations of the algorithm rely on the computation of Eigenfaces derived from an image dataset; however, a critical limitation arises due to the requirement that the face images within the dataset and unseen faces must be captured under uniform lighting conditions. Variations in lighting between a person's dataset images and unseen faces significantly impact the algorithm's accuracy, thereby limiting its practical applicability. To address this limitation, the Weber Local Descriptor (WLD) is introduced as a local descriptor employed during the algorithm's training and recognition phase. The application of WLD facilitates the normalization of light levels across images, effectively mitigating the adverse effects of varying lighting conditions on the performance of the Eigenface algorithm. The extended Yale B dataset, which has 2,414 images with varying lighting conditions, is used to test the algorithm’s accuracy. Using stratified k-fold cross-validation with 10 splits, the original algorithm achieved 77.84% accuracy, while the enhanced Eigenface algorithm achieved 99.79% accuracy. This enhancement significantly improved the Eigenface algorithm's ability to recognize faces under varying lighting conditions, making it more practical in an attendance management system.