Real-Time Transformer-Based Perception for Intelligent Transportation System

DOI

10.2991/978-94-6239-616-6_67How to use a DOI?

Keywords

component; formatting; style; styling; insert

Abstract

Vehicle detection in real-time accurately is a vital component in modern Intelligent Transport Systems (ITS) for eliminating urban traffic congestion. While there are many deep learning models often a trade-off between accuracy and inference speed exists. This paper shows a complete high performing baseline system in real-time traffic analysis which is built using RT-DETR (Real-Time Detection Transformer).[1] We outlined a detailed data to deployment pipeline starting from the original UA-DETRAC dataset. A key contribution of this work is a robust preprocessing pipeline that transforms the dataset’s complex sequence-based XML annotations into a streamlined YOLO-formatted dataset consisting of 82,085 samples. The processed dataset then was used to fine-tune the RT-DETR model by transfer learning on a local workstation equipped with a consumer grade NVIDIA RTX series GPU. The results demonstrate that the fine- tuned RT-DETR-L model achieves a Mean Average Precision (mAP) of 78.9% at 54 FPS, outperforming baseline YOlOv8 architectures in complex traffic scenar- ios including a virtual line crossing algorithm using the model’s built in tracking features to IDs and accurately count vehicles. Our work verifies that deploying advanced real time transformer models on non-enterprise hardwire produces a framework that serves as foundational baseline for advanced ITS applications.[2]

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - D. Bharadwaja
AU  - Karimireddy Sai Sandeep Reddy
AU  - Mudrageda Bhargava Phani Sriram
AU  - Dhruv Bavaria
PY  - 2026
DA  - 2026/03/31
TI  - Real-Time Transformer-Based Perception for Intelligent Transportation System
BT  - Proceedings of the International Conference on Artificial Intelligence and Secure Data Analytics (ICAISDA 2025)
PB  - Atlantis Press
SP  - 897
EP  - 909
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-616-6_67
DO  - 10.2991/978-94-6239-616-6_67
ID  - Bharadwaja2026
ER  -