Enhancing Autonomous Vehicle Navigation: Traffic Police Hand Gesture Recognition for Self-Driving Cars in India Using MoveNet Thunder

Ippili Rahul, Saravanan Santhanam

2025

Abstract

The Traffic Police Hand Gesture Recognition System for autonomous vehicles implements the TensorFlow’s MoveNet Thunder model. The system detects three hand signals including 'Stop' and 'Turn Left' and 'Move Forward' since such movements correspond to standard traffic police actions in India. Our custom database included eight thousand gestures’ images recorded under diverse circumstances. An architecture of dense and dropout layers within our network enables both accurate performance while keeping the network protected from overfitting conditions. The Haar cascades face detection system enables live officer identification before camera recording of gestures that occur within the field of view starts. Minor mistakes occurred between similar gestures despite the system achieving a 89% success rate. System evaluation using Carla simulator was conducted under two conditions: first with environmental conditions enabled and second without environmental conditions enabled. The created prototype proves useful as an effective element that combines safety features with operational efficiency for autonomous vehicle navigation systems in controlled traffic environments.

Paper Citation

in Harvard Style

Rahul I. and Santhanam S. (2025). Enhancing Autonomous Vehicle Navigation: Traffic Police Hand Gesture Recognition for Self-Driving Cars in India Using MoveNet Thunder. In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25; ISBN 978-989-758-777-1, SciTePress, pages 473-479. DOI: 10.5220/0013931600004919

in Bibtex Style

@conference{icrdicct`2525,
author={Ippili Rahul and Saravanan Santhanam},
title={Enhancing Autonomous Vehicle Navigation: Traffic Police Hand Gesture Recognition for Self-Driving Cars in India Using MoveNet Thunder},
booktitle={Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25},
year={2025},
pages={473-479},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013931600004919},
isbn={978-989-758-777-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25
TI - Enhancing Autonomous Vehicle Navigation: Traffic Police Hand Gesture Recognition for Self-Driving Cars in India Using MoveNet Thunder
SN - 978-989-758-777-1
AU - Rahul I.
AU - Santhanam S.
PY - 2025
SP - 473
EP - 479
DO - 10.5220/0013931600004919
PB - SciTePress