ABSTRACT
ABSTRACT: Face recognition is transforming the way people are interacting with machines. Earlier it was used in specific domains like law enforcement but with extensive research being done in this field, it is being extended to various applications like automatic face tagging in social media, surveillance systems in airports, theaters and so on. Local feature detection and description is gaining a lot of significance in the face recognition community. Extensive research on SURF and SIFT descriptors have found widespread application. Key points matched by SURF and triangulated using Delaunay Triangulation boosts the interest points detected. Other modern techniques like machine learning and deep learning require huge amount of training data and computational capabilities which sometime becomes a limitation of its usage. In contrast, hand-crafted models like SURF, SIFT overcomes the requirement of training data and computing power. The pipeline proposed in the paper reduces the average computational power and increases accuracy.
