iNFRA’s AI Vehicle Classification System Featured in the Civil Insight Series by the Civil Engineering Committee of EIT
iNFRA Co., Ltd. is proud to announce that its article, “A New Era of Traffic Data Collection with Artificial Intelligence,” has been featured in Civil Insight Series Content (Ep.46 DIT-CACE/004) by the Civil Engineering Committee of the Engineering Institute of Thailand under H.M. The King’s Patronage (EIT). This recognition reflects iNFRA’s expertise in applying Artificial Intelligence (AI) to support traffic engineering and infrastructure development.
A New Era of Traffic Data Collection with Artificial Intelligence
Traffic volume surveys provide fundamental data for road design, infrastructure planning, traffic management, and Smart City development. Traditionally, traffic surveys relied on personnel manually counting vehicles from video recordings or at survey sites, a process that is time-consuming, costly, and susceptible to human error.
Today, Artificial Intelligence (AI) combined with Computer Vision is transforming traffic data collection by enabling faster, more accurate, and real-time detection, analysis, and data processing.
What is the AI Vehicle Classification System?
The AI Vehicle Classification System is a Deep Learning-based solution that analyzes video footage to automatically detect, track, and classify vehicles while processing traffic data in real time.
The system integrates several core technologies, including:
- YOLOv10 for Object Detection
- Multi-Object Tracking (ByteTrack / OC-SORT)
- Vehicle Classification
- Video Analytics
Why is Multi-Object Tracking Important?
Object Detection alone cannot determine whether a detected vehicle is the same one appearing across consecutive video frames or a different vehicle. Without an effective tracking mechanism, a single vehicle may be counted multiple times as it moves through the video.
Multi-Object Tracking addresses this challenge by assigning a unique Object ID to each vehicle, continuously tracking its movement, preventing duplicate counting, and improving the accuracy of traffic data for engineering analysis.
In simple terms, Detection enables the system to “see” an object, while Tracking enables it to “understand” the object’s movement. Together, these technologies provide reliable data for engineering applications.
System Capabilities
The AI Vehicle Classification System is capable of:
- Classifying 13 vehicle categories
- Operating under various lighting conditions
- Performing real-time traffic analysis
- Achieving an accuracy of over 91%
Engineering Applications
The traffic data generated by the AI Vehicle Classification System can support a wide range of engineering and infrastructure planning applications, including:
- Traffic Survey
- Highway Engineering
- Pavement Design
- Traffic Impact Assessment (TIA)
- Intelligent Transportation Systems (ITS)
- Smart City
- Transportation Planning
- Digital Twin
The publication of this article in Civil Insight Series Content by the Civil Engineering Committee of the Engineering Institute of Thailand under H.M. The King’s Patronage (EIT) marks another milestone for iNFRA. It reflects the company’s ongoing commitment to advancing engineering knowledge and technological innovation to support the development of transportation systems and sustainable infrastructure.
Read the Full Article : https://www.facebook.com/share/p/18vUrxaqP4/
