Comparative Analysis of ANNs and SNNs for Road Sign Detection in Autonomous Driving

Abstract

Understanding and accurately interpreting road signs are essential capabilities for autonomous driving systems, ensuring safety and efficient navigation. This paper compares two neural network models: Artificial Neural Networks (ANNs) and Spiking Neural Networks (SNNs), focusing on their performance and potential for implementation on Field-Programmable Gate Arrays (FPGAs). Both models were trained on a diverse collection of road sign data, with their performance evaluated based on metrics such as accuracy, precision, and energy efficiency. The findings demonstrate that while both models achieve high accuracy, SNNs excel in energy efficiency, making them a more practical choice for FPGA-based systems. These results emphasize the promise of SNNs as a competitive and energy-efficient solution for real-time traffic sign recognition in autonomous driving applications.