On Matching Faces with Alterations due to Plastic Surgery and Disguise

Abstract

Plastic surgery and disguise variations are two of the most challenging co-variates of face recognition. The stateof-art deep learning models are not sufficiently successful due to the availability of limited training samples. In this paper, a novel framework is proposed which transfers fundamental visual features learnt from a generic image dataset to supplement a supervised face recognition model. The proposed algorithm combines off-the-shelf supervised classifier and a generic, task independent network which encodes information related to basic visual cues such as color, shape, and texture. Experiments are performed on IIITD plastic surgery face dataset and Disguised Faces in the Wild (DFW) dataset. Results showcase that the proposed algorithm achieves state of the art results on both the datasets. Specifically on the DFW database, the proposed algorithm yields over 87% verification accuracy at 1% false accept rate which is 53.8% better than baseline results computed using VGGFace.

Impact of Digital Face Beautification in Biometrics

“…These surgery changes pose a challenge to automatic FR technology [3]. Although facial plastic surgery is usually employed for cosmetic and scars treatment to improve the person's appearance, it might also be used by criminals to 'manipulate' their facial identity with the intent to deceive FR systems [11], [12].…”

On the Robustness of Face Recognition Algorithms Against Attacks and Bias

“…Marsico et al (2015) proposed region-based strategies for face recognition under variations due to plastic surgery. Suri et al (2018) proposed a COST framework (COlour, Shape, and Texture) for matching pre and post plastic surgery face images. For detecting faces which have undergone plastic surgery, a Multiple Projective Dictionary Learning based technique has been proposed, followed by a face verification pipeline utilizing the information from the altered and non-altered regions (Kohli, Yadav, and Noore 2015).…”

Impact and Detection of Facial Beautification in Face Recognition: An Overview

“…Notably, over the past years, face recognition performance has significantly improved. Specifically, while early performance rates comprised 34.1% GMR at 0.1% FMR (associated to 2-D Log Polar Gabor Transform (GNN) [20]), recent deep learning approaches have significantly increased performance rates to 91.75% R-1 and ∼90% GMR at 0.1% FMR (associated to a method introduced by Suri et al [32]). Further, it can be observed that many approaches, which were designed to be resilient to plastic surgery, process face images in a patch-wise manner, also referred to as ''part-wise'', ''image block-wise'' or ''sub-region-wise'', e.g., [22], [23], [26], [28].…”