Traffic Sign Recognition and Speech-Based Feedback Generation System for Driver Assistance
“Deep learning embedded vision system running on edge hardware to recognize road sign typography under occlusions and provide low-latency spoken alerts to drivers.”
Project Scope & Details
A robust advanced driver-assistance system (ADAS) integrating convolutional neural networks for real-time traffic sign detection, classification under varying lighting and adverse weather conditions, and synthetic speech synthesis for non-intrusive auditory alerts.
Edge Deep Vision Convolutional Pipeline
Lightweight neural architecture detecting and classifying regulatory, warning, and informational signs at 120 FPS.
Real-Time Driver Warning HUD
Synthesizes low-distraction visual and auditory alerts for speed limits, school zones, and sudden road hazards.
Adverse Weather & Occlusion Robustness
Trained on challenging Indian driving conditions, including heavy monsoon rain, fog, and tree canopy occlusion.
Embedded Low-Power In-Cabin Deployment
Operates within a 5W power budget on edge AI microcomputers (Jetson Nano/Orin) for broad fleet retrofitting.
Traffic Sign Detection & Feedback Architecture
A Traffic Sign Recognition and Speech-Based Feedback Generation System to Assist Drivers
Authors: Himangshu Sarma, Navanath Saharia
- Dr. Himangshu Sarma
- Dr. Navanath Saharia