Real-Time Traffic Sign Recognition for Driver Assistance Using HSV Color Space and Lightweight CNN Models

This research paper presents a real-time traffic sign recognition (TSR) system developed to enhance driver safety by precisely detecting and classifying road signs under various lighting and occlusion conditions. The system merges classical image processing methods (such as HSV color segmentation and shape-based feature extraction) with a lightweight convolutional neural network (CNN) classifier. The robustness of the system is bolstered by an extensive data-augmentation pipeline that incorporates geometric transformations, photometric alterations, and synthetic occlusions. For deployment, we discuss a low-power embedded hardware option and acceleration techniques to fulfill real-time constraints. Experimental evaluations on the German Traffic Sign Recognition Benchmark (GTSRB) and additional in-house test sets indicate high classification accuracy (greater than 98% on clean data) with latency suitable for in-vehicle use (target ≥ 15 fps). The paper provides a detailed account of the methodology, implementation, results, and design trade-offs, serving as a practical guide for production-ready TSR systems.