COMSATS Face

Abstract

The human face can appear different depending on the circumstances because of its flexibility and three-dimensional structure. Researchers are facing several obstacles relating to face poses, illumination, facial expressions, head direction, occlusion, hairdo, etc. in the process of developing dependable and efficient algorithms for face detection, face identification, and face expression analysis. To determine the algorithms' effectiveness, they need to be evaluated against a certain set of face image/database benchmarks. This work introduces a dataset of multiple-pose facial photographs. Eight hundred fifty photos from 50 people in 17 distinct stances are included in the collection (0°, 5°, 10°, 15°, 20°, 25°, 30°, 35°, 55°, -5°, -10°, -15°, -20°, -25°, -30°, -35°, -55°). Three distinct lighting conditions are also included in the dataset. Eight resolutions (144 × 256, 200 × 200, 100 × 100, 70 × 70, 50 × 50, 40 × 40, 20 × 20 and 10 × 10 pixels) are available for the dataset. This dataset's facial image content can provide insight on the effectiveness and resilience of upcoming face detection and recognition systems. Additionally, based on the suggested face database, a comparison study of two face recognition methods, such as PAL and PCA, is performed.