International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 05 | May 2025
p-ISSN: 2395-0072
www.irjet.net
RoadEye-A YOLOv12-Based Approach for Real-Time Road Pothole Detection Dr. Raghavendra K1, Sakshi Shankar Gumaste2, Sanjana Singh3, Tejas K 4 ,Varsha N 5 1Associate Professor,Dept. of CSE Engineering, Jyothy Institute of Technology,Bangalore,India 2Dept. of CSE ,Jyothy Institute of Technology,Bangalore,India 3
Dept. of CSE ,Jyothy Institute of Technology,Bangalore,India
4Dept. of CSE ,Jyothy Institute of Technology,Bangalore,India 5Dept. of CSE ,Jyothy Institute of Technology,Bangalore,India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Road surface disruptions such as potholes
to a central server where it undergoes preprocessing and analytical assessment to ascertain the specific type of surface disruption [4].
critically affect traffic flow, vehicle safety, and maintenance strategies. This study introduces an efficient second-order hyperbolic traffic model that accounts for pothole width, driver reaction, and time headway, improving traffic evolution prediction compared to traditional models. A lightweight sensing system has been created to enable real-time detection using vibration signals, spatio-temporal trajectory fusion, and ultrasonic sensors applicable to both paved and unpaved roads. This system achieves 94% accuracy and operates in real-time utilizing NB-IoT and embedded signal processing. Furthermore, an analysis of machine learning and IoT-based frameworks reveals scalable methods for automated monitoring of road conditions. This study connects traffic modeling with advanced sensing for proactive maintenance of roads in smart cities.
The core methodology involves deploying the ultrasonic sensor system using Arduino hardware for real-time data collection, followed by software-based processing and classification. The proposed framework enhances road maintenance by ensuring early detection and cost-efficient intervention. Section II of the paper discusses existing literature on RSD detection and infrastructure monitoring. Section III outlines the proposed system architecture including data acquisition, processing techniques, and anomaly classification algorithms. Section V presents a comprehensive evaluation of the system’s effectiveness, along with resource efficiency and operational feasibility. The paper concludes in Section VI with future recommendations and potential directions for scaling this solution to urban-level deployments.
Key Words: Transportation Safety Traffic Modeling, Pothole Detection, Smart Cities, Real-Time Sensing, YOLOv12, Road Surface Monitoring
2. LITERATURE REVIEW
1.INTRODUCTION
2.1. Categories of Traffic Flow Modeling
Modern road infrastructure is essential for fostering national development by facilitating the efficient movement of goods and people. However, rapid urban growth and a rising number of vehicles on the road have led to significant traffic congestion and an increased risk of road accidents. One major factor contributing to these issues is the occurrence of Road Surface Disruptions (RSDs) [3], which include potholes, cracks, and other deformities that adversely affect road conditions.
Traffic flow analysis plays a crucial role in managing road systems. Depending on the detail and scale, traffic models are commonly split into three categories: •
o Focus on individual vehicle dynamics such as velocity, acceleration, and spacing. o Studies emphasize driver behaviors like headway maintenance and braking responses [5].
This research proposes an effective and cost-efficient method for detecting surface irregularities on roads by using an ultrasonic sensor-based system. The approach is designed to be applicable to both paved and unpaved roads, allowing for wider usability and real-time feedback [2]. Conventional detection methods such as vibration-based, laser-based, and vision-based techniques have demonstrated shortcomings in terms of precision, scalability, or cost-effectiveness. When mounted on a vehicle, the ultrasonic sensor actively evaluates the road surface to identify irregularities. Once an irregularity is detected, the system transmits the sensor data
© 2025, IRJET
|
Impact Factor value: 8.315
Microscopic Models:
•
Macroscopic Models: o Analyze aggregate traffic behavior using variables like density and flow. o Introduced the LWR model as the first macroscopic model [6], [9].
•
Mesoscopic Models: o Combine individual and group behaviors to balance detail and computational efficiency
|
ISO 9001:2008 Certified Journal
|
Page 304