Traffic Sign Recognition Using Hybrid Deep Ensemble Learning for Advanced Driving Assistance Systems

Abstract

Automakers are working on merging Advanced Driving Assistance Systems (ADAS) with artificial intelligence and computer vision to reduce the number of fatalities and accidents brought on by distracted driving. One of the primary criteria for autonomous vehicles and the majority of ADAS is the capacity to perceive and comprehend all static and moving objects surrounding a vehicle in a variety of driving and environmental situations. Artificial intelligence may fulfill the current promise to supply safe ADAS in contemporary vehicles (AI). To support ADAS, this article demonstrated automated traffic sign identification for the Indian Traffic Scenario. The traffic signs are categorized into their superclass and subclasses. Deep learning is employed for feature extraction and ensemble learning for classification. The testing results using databases of Indian traffic signs demonstrated that the proposed method performed on par with state-of-the-art techniques and that the processing effectiveness of the entire recognition process was also increased.