I-SOCIAL-DB: A labeled database of images collected from websites and social media for Iris recognition

People upload daily a huge number of portrait face pictures on websites and social media, which can be processed using biometric systems based the characteristics to perform an automatic recognition individuals. However, performance approaches limited by negative factors as aging, occlusions, rotations, uncontrolled expressions. Nevertheless, constantly increasing quality resolution uploaded media could permit overcome these problems improve robustness methods enabling analysis additional traits, like iris. To point attention research community possible use iris-based techniques for images we present public image dataset called I-SOCIAL-DB (Iris Social Database). This is composed 3,286 ocular regions, extracted from 1,643 high-resolution 400 individuals, collected websites. For each region, human expert coordinates circles approximating inner outer iris boundaries performed pixelwise segmentation contours, reflections. first collection one biggest collections manually segmented in literature. In this paper, also qualitative samples, set testing protocols figures merit, benchmark results achieved publicly available algorithms. We hope that initiative give new test tool community, aiming stimulate studies challenging field.