OBJECT DETECTION AND CLASSIFICATION FOR AUTONOMOUS VEHICLE USING VISION TRANSFORMER

Dr.B.Vijayalakshmi , Dr.I.Gethzi Ahila Poornima , Dr.M.Swarna Sudha

Vision Transformer, Autonomous Vehicles, Object Detection, Self-Attention, Deep Learning, LabelMe Dataset, COCO Labels, CNN, Image Classification

Abstract: This paper presents an object detection and classification system for autonomous vehicles using Vision Transformers (ViTs). Object detection is a critical component in autonomous driving, enabling the vehicle to perceive its surroundings accurately and make informed decisions. Traditional methods largely rely on Convolutional Neural Networks (CNNs), which have limitations in capturing long-range dependencies within image data. Vision Transformers, inspired by the success of transformer models in natural language processing, model global relationships in visual data. This work implements a ViT-based approach using PyTorch, trained and evaluated on the LabelMe-12-50k dataset with COCO label taxonomy. The system detects and classifies vehicles, pedestrians, and traffic signs in real-time. The ViT model achieved 89.7% accuracy, 88.3% precision, 87.9% recall, and an F1 score of 88.1%, outperforming CNN baselines across all metrics and validating the effectiveness of Vision Transformers for autonomous vehicle perception.

Published Paper   E-Certificate

"OBJECT DETECTION AND CLASSIFICATION FOR AUTONOMOUS VEHICLE USING VISION TRANSFORMER", JETNR - JOURNAL OF EMERGING TRENDS AND NOVEL RESEARCH (www.JETNR.org), ISSN:2984-9276, Vol.4, Issue 6, page no.a388-a393, June-2026, Available :https://rjpn.org/JETNR/papers/JETNR2606052.pdf

Volume 4 Issue 6, June-2026

Pages : a388-a393

Paper Reg. ID: JETNR_234998

Published Paper Id: JETNR2606052

Downloads: 00072

Research Area: Science and Technology

Country: Rajapalayam, Tamil Nadu, India

Published Paper PDF: https://rjpn.org/JETNR/papers/JETNR2606052

Published Paper URL: https://rjpn.org/JETNR/viewpaperforall?paper=JETNR2606052

ISSN: 2984-9276 | IMPACT FACTOR: 9.87 Calculated By Google Scholar | ESTD YEAR: 2023

An International Scholarly Open Access Journal, Peer-Reviewed, Refereed Journal Impact Factor 9.87 Calculate by Google Scholar and Semantic Scholar | AI-Powered Research Tool, Multidisciplinary, Monthly, Multilanguage Journal Indexing in All Major Database & Metadata, Citation Generator

Publisher: RJPN (IJPublication) Janvi Wave

This work is licensed under a Creative Commons Attribution 4.0 International License and The Open Definition