Face Extraction and Recognition from Public Images Using HIPI

Andrey Bogdanchikov, Darmen Kariboz, Meraryslan Meraliyev

2018

Closed access Metadata (CSL-JSON)
DOI 10.1109/icecco.2018.8634718
Data confidence 85 high
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

Social networking services with public data are widely used nowadays. Billions of images uploaded to the internet each day over the world. This paper proposes the idea of a system which is currently being developed. The system collects images from public sources by some specific criteria, applies face detection and recognition algorithms on collected images, and provides the results in readable form. The Apache Hadoop library is used to increase performance of the system, and images are downloaded by the help of HIPI library. For testing purposes Labeled Faces in the Wild benchmark is used as a database for images containing over 13233 images of 5749 identities. Each image is jpg format in 250*250 resolutions. Results were more than good after testing the system with this database for face detection and recognition.