MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202621048300 A) filed by Symbiosis International Deemed University on April 15, 2026, for Automatic Traffic Sign Detection And Recognition System For Smart Vehicles Using Deep Learning.
Inventors include Dr. Gagandeep Kaur; Dr. Latika Pinjarkar; Fuzail Sheikh; and Atharva Vairagade.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: ABSTRACT AUTOMATIC TRAFFIC SIGN DETECTION AND RECOGNITION SYSTEM FOR SMART VEHICLES USING DEEP LEARNING The present invention discloses an automatic traffic sign detection and recognition system (100) for smart vehicles comprising an image acquisition module (110), an image preprocessing module (120) with haze removal unit (HRU-Net) (122), a traffic sign detection module (130) employing YOLOv8 architecture (132) enhanced with Coordinate Attention (CA) mechanism (134) and Bi-directional Feature Pyramid Network (BiFPN) (136), a traffic sign classification module (140) utilizing hybrid CNN-Transformer architecture (142), a rule-based validation module (150), a vehicle control interface module (160), and a central processing unit (170). The system achieves real-time detection accuracy exceeding 97% with inference time of 22 milliseconds per frame, and classification accuracy exceeding 98%. The invention addresses limitations of prior art by providing robust performance under adverse environmental conditions, multi-scale feature fusion for detecting signs at various distances, and explainable decision-making through rule-based validation for enhanced safety in autonomous driving applications. [
Disclaimer: Curated by HT Syndication.
