International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 11 Issue: 05 | May 2024
www.irjet.net
p-ISSN: 2395-0072
A NOVEL APPROACH FOR AN EFFICIENT MANHOLE VISUAL INSPECTION SYSTEM USING DEEP LEARNING TECHNIQUES 1Dr. T. Amalraj Victorie, 2M. Vasuki, 3Suriya.P
Professor, Department of Master of Computer Application, Sri Manakula Vinayagar Engineering College Puducherry-605 107,India. 2Associate Professor, Department of Master of Computer Application, Sri Manakula Vinayagar Engineering College Puducherry-605 107,India. 3PG Student, Department of Master of Computer Application, Sri Manakula Vinayagar Engineering College Puducherry-605 107,India. ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - A manhole serves as a covered access point in
Our proposed system utilizes deep neural networks for image analysis, enabling automatic detection and classification of manhole defects based on severity and type. Through training on a diverse dataset of manhole images, the deep learning model learns to identify various anomalies, including cracks, structural damage, and surface deterioration. Furthermore, the system integrates real-time image processing capabilities, enabling prompt analysis of manhole images captured by inspection cameras or mobile devices. This facilitates timely detection and assessment of defects, allowing for proactive maintenance interventions and minimizing the risk of accidents and infrastructure damage.
streets or public areas, allowing entry to underground utility or maintenance vaults. Typically sealed with a lid, these openings facilitate inspection, upkeep, or repair of various systems like sewers, telecommunications, or gas lines. Given their public location, securing manhole covers is crucial to prevent accidents, as damaged or missing ones pose significant risks to pedestrians, cyclists, and motorists. The deterioration of manhole covers raises concerns about traffic accidents, urging the need for a more efficient inspection method. Traditional manual observation faces challenges like labour shortages and ethical issues. Meanwhile, utilizing image processing algorithms to detect open or damaged manholes encounters difficulties due to varying image quality, complex backgrounds, and changing environmental conditions. To address these challenges, a project proposes an automated system architecture leveraging deep learning models to replace manual inspections. This involves developing a deep learning model capable of analysing images of manhole covers captured through CCTV footage.
This paper presents the architecture and implementation details of our manhole visual inspection system, highlighting its key components such as deep learning models, image processing techniques, and data acquisition methods. Additionally, empirical results and performance evaluations are provided to demonstrate the effectiveness and efficiency of the system in accurately detecting and classifying manhole defects. Overall, our research contributes to advancing infrastructure maintenance practices by introducing a reliable and automated approach to manhole visual inspection. By leveraging the power of deep learning, we aim to enhance the safety and reliability of transportation networks while reducing the burden on manual inspection processes.
Key Words: Manhole covers, Deep learning models, Infrastructure surveillance.
maintenance,
urban
safety,
CCTV
1.INTRODUCTION: Ensuring safe and efficient transportation systems necessitates meticulous maintenance of road infrastructure, with road holes posing significant challenges due to their potential to cause accidents and disrupt traffic flow. Traditional manual visual assessments for road-hole inspection are labor-intensive, time-consuming, and prone to human error. To overcome these limitations, this study proposes an innovative manhole visual inspection system leveraging deep learning technology. By harnessing advanced deep learning algorithms, our system aims to automate and optimize the inspection process, thereby improving efficiency and accuracy.
© 2024, IRJET
|
Impact Factor value: 8.226
2. LITERATURE SURVEY: 1.
"Automatic Inspection of Sewer Systems Using Deep Learning" by Tiefeng Liu, Yuzhuo Ren, and Weiliang Jin (2018):
|
This paper proposes a method for automatically detecting and classifying defects in sewer systems using deep learning techniques. The authors utilize convolutional neural networks (CNNs) for feature extraction and defect classification. The study demonstrates the effectiveness of deep
ISO 9001:2008 Certified Journal
|
Page 1084