Low Light Face Detection System

Low Light Face Detection System

G. Ananthi, T. Balaji, M. Pugalenthi, M. Rajkumar
Copyright: © 2023 |Pages: 16
DOI: 10.4018/978-1-6684-7216-3.ch008
OnDemand:
(Individual Chapters)
Available
$33.75
List Price: $37.50
Current Special Offers
10% Discount:-$3.75
TOTAL SAVINGS: $3.75

Abstract

Detection of faces from low light image is challenging. Such images often suffer from poor contrast, poor intensity, and high noise make it challenging to gather information and identify people or objects. The proposed method uses a deep learning-based approach to enhance the images by capturing multiple exposures and combining them to generate a high-quality image. Max-margin object detection (MMOD) human face detector, a face detection algorithm is used to identify and enhance the faces in the images. MMOD algorithm is a deep learning-based object detection approach which accurately detects human faces in images. By identifying and enhancing the faces in the images, they are made more visible and recognizable, even in low light conditions. The experimental results of this proposed approach demonstrate its effectiveness in enhancing face visibility and raising the quality of low-light images. This method has practical applications in areas such as security surveillance, where capturing high-quality images under low light conditions is critical.
Chapter Preview
Top

Background And Literature Review

Yang et al. (2020) focused on finding objects or faces in low-light conditions brought on by inclement weather (haze, rain). In order to offer a more complete analysis and a reasonable comparison, they offered three benchmark sets with annotated objects/ faces that were captured in foggy, rainy, and low-light conditions. This is the first and best initiative of its kind right now, to the best of our knowledge, in order to elicit a thorough investigation into whether and how low-level vision approaches might aid high- level automatic visual identification in many contexts. By cascading current enhancement and detection models, they provided baseline results while emphasizing the new data's extreme difficulty and the need for further technical development. There has been a lot of research on face detection (Zhou et al., 2018). Extreme posture, lighting, low resolution, and small scales are just a few of the challenges faced by face detectors that have been analyzed in the prior work. The authors investigated how low-resolution photos with varying levels of befog, noise, and contrast degrade their performance. Their research demonstrates that hand-built face detectors and those based on deep learning are insufficiently resistant to subpar images. It inspires researchers to develop more trustworthy face-identification algorithms for dark images.

Complete Chapter List

Search this Book:
Reset