Log In Sign Up

A study of Biases in Common Face Recognition Datasets

by   Kanishk Rath, et al.

The performance of commonly used facial recognition datasets have been studied when applied on a racially diverse population. A need for racial diversity in the training data set has been observed to improve the outputs of the models. Further, impact of Covid-19 pandemic on facial recognition technologies in terms of Mask detection models and corresponding datasets have been studied. Finally, a GAN based tool has been used for modifying image using sentences to obtain racial or gender diverse outcome. Potential implications of GAN based tools for such image manipulation, with potential solution to avoid misuse of such software have been discussed.


page 5

page 7


Towards NIR-VIS Masked Face Recognition

Near-infrared to visible (NIR-VIS) face recognition is the most common c...

Can GAN-induced Attribute Manipulations Impact Face Recognition?

Impact due to demographic factors such as age, sex, race, etc., has been...

Masked Face Recognition for Secure Authentication

With the recent world-wide COVID-19 pandemic, using face masks have beco...

Facial Expressions as a Vulnerability in Face Recognition

This work explores facial expression bias as a security vulnerability of...

An Examination of Bias of Facial Analysis based BMI Prediction Models

Obesity is one of the most important public health problems that the wor...

Real Time System for Facial Analysis

In this paper we describe the anatomy of a real-time facial analysis sys...

Facial transformations of ancient portraits: the face of Caesar

Some software solutions used to obtain the facial transformations can he...