Research Article
An Offline Image Auditing System for Legacy Meter Reading
Systems in Developing Countries: A Machine Learning Approach
Natasha Nigar ,
1
Hafiz Muhammad Faisal ,
2
Muhammad Kashif Shahzad,
3
Shahid Islam ,
1
and Olukayode Oki
4
1
Department of Computer Science (RCET), University of Engineering and Technology, Lahore, Pakistan
2
School of Engineering and Design, Technical University of Munich, Munich, Germany
3
Power Information Technology Company (PITC), Ministry of Energy, Power Division, Goverment of Pakistan, Lahore, Pakistan
4
Department of Information Technology, Walter Sisulu University, Mthatha, South Africa
Correspondence should be addressed to Natasha Nigar; natasha@uet.edu.pk
Received 29 September 2022; Revised 11 November 2022; Accepted 14 November 2022; Published 25 November 2022
Academic Editor: Raid Al•Nima
Copyright©2022NatashaNigaretal.TisisanopenaccessarticledistributedundertheCreativeCommonsAttributionLicense,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Te developing countries are challenged with overbilling and underbilling, due to manual meter reading, which results in
consumer dissatisfaction and loss of revenue. Te existing automated meter reading (AMR) solutions are expensive; hence,
sample•based manual snap auditing systems are introduced to control such meter reading inaccuracies. In these systems, the
meter reader, besides reading, also collects meter images, which are used to manually audit the meter’s accuracy. Although such
systems are inexpensive, they are limited in their ability to be sustainable and ensure 100% accurate meter readings. In this paper, a
novel ofine optical character recognition (OCR) system•based Snap Audit system is proposed and tested foritsefcient and real•
time 100% accurate meter reading capabilities. Te experimental results on 5,000 real•world instances show that the proposed
approach processes an image in 0.05 seconds with 94% accuracy. Moreover, the developed approach is evaluated with four state•
of•the•art algorithms: region convolution neural network (RCNN), nanonets, Fast•OCR, and PyTesseract. Te results provide
evidence that our new system design along with novel approach is more robust and efcient as compared to existing algorithms
by 43.6%.
1. Introduction
In general, employees from utility companies (electricity,
gas, and water) record the consumption manually by
walking from house to house and/or building to building on
monthly basis. Te reading is manually entered in the
mobile•based meter reading systems along with meter dial
pictures 1–3] for later verifcation and audit purposes due to
extended likelihood of error 4–6]. Te manual audit of
meter readings collected by the feld staf on a monthly basis
is an expensive and monotonous task in terms of human
efort and time 7] and has low accuracy. Te utility com•
panies adopt stratifed sampling to make a general con•
clusion about the population, which is based on monthly
meter readings. Moreover, manual evaluation of a large
number of images may result in negligence of errors.
Besides all benefts and extended analysis support ofered
by AMR meters 8], upfront cost, infrastructure cost, and
technical capacity to sustainable operation are limiting
factors for its rollout in the developing countries and the
meter reading process is manual. Hence, there is a need to
provide a technology solution that provides automation in
the existing system without any fnancial burden or ad•
vanced training to operate and sustain while ensuring 100%
accurate meter reading. Over the last decade, the trend to
digitize paper•based documents has emerged 9]. Te aim is
to make these documents fully searchable, accessible, and
processable in digital form. In this regard, OCR technology
has gathered researchers’ attention 10]. OCR is the me•
chanical or electronic translation of scanned images of
handwritten, typewritten, or printed text into machine•
encoded text 11]. Te OCR ideally returns the same output
Hindawi
Journal of Electrical and Computer Engineering
Volume 2022, Article ID 4543530, 10 pages
https://doi.org/10.1155/2022/4543530