Occluded face detection, face in niqab dataset

Abstract

The detection of faces with occlusion is one of the remaining challenges in the field of computer vision. Despite the high performance of current advanced face detection algorithms, occluded faces are under hot research and is still requires more investigation. There is a need of rich dataset for occluded faces to be used for enriching the training of cnn-deep learning models in order to boost the performance of face detection for occluded faces as in face in niqab. In this paper a dataset of occluded faces named faces in niqab is proposed. An experimental results indicated the poor performance of state-of-the-art face detection algorithms when tested in our dataset.