Computer Vision based Detection of Partially Occluded Faces

In today’s world, security has gained utmost significance in every walk of life. With the recent advancements in image and video analytics, emphasis has been towards developing enhanced surveillance systems which perform complex tasks that include automated security incident detection, tracking and analysis in real time. The primary objective of this paper is to automatically detect the presence of any masked or occluded face in real time. A robust technique based on pivotal facial points has been developed. The paper discusses in detail how the pivotal points are observed extracted are used in discovering masked faces in real time. Analysis of this algorithm’s performance on test data sets gives positive insights for further enhancements towards occluded face detection in real time surveillance.