D-LORD: DYSL-AI Database for Low-Resolution Disguised Face Recognition

Face recognition in a low-resolution video stream captured from a surveillance camera is a challenging problem. The problem becomes even more complicated when the subjects appearing in the video wear disguise artifacts to hide their identity or try to impersonate someone. The lack of labeled datasets restricts the current research on low-resolution face recognition systems under disguise. With this paper, we propose a large-scale database, D-LORD, that will facilitate the research on face recognition. The proposed D-LORD dataset includes high-resolution mugshot images of 2,100 individuals and 14,098 low-resolution surveillance videos, collectively containing over 1.2 million frames. Each frame in the dataset has been annotated with five facial keypoints and a single bounding box for each face. In the videos, subjects’ faces are occluded by various disguise artifacts, such as face masks, sunglasses, wigs, hats, and monkey caps. To the best of our knowledge, D-LORD is the first database to address the complex problem of low-resolution face recognition with disguise variations. We also establish the benchmark results of several state-of-the-art face detectors, frame selection algorithms, face restoration, and face verification algorithms using well-structured experimental protocols on the D-LORD dataset. The research findings indicate that the Genuine Acceptance Rate (GAR) at 1% False Acceptance Rate (FAR) varies between 86.44% and 49.45% across different disguises and distances. The dataset is publicly available to the research community at https://dyslai.org/datasets/D-LORD/.

I. Introduction

Low-resolution face recognition is a critical task in many security-based applications, where the gallery is usually a High Resolution (HR) image, and the probe is a Low Resolution (LR) image. Images or videos captured by surveillance cameras are usually affected by non-ideal conditions such as poor resolution, blurriness, non-uniform illumination conditions, and non-frontal face position. It is challenging to construct robust feature embeddings due to limited information content in LR probe images. The recent literature reports a significant drop in the performance of automated face recognition models when tested on LR inputs compared to HR probes [1]. We claim that real-world LR images pose greater challenges to existing face detectors and face recognition algorithms as compared to synthetic images downscaled from HR images [2].