Abstract
This paper represents a study for the realization of a system based on Artificial Intelligence, which allows the recognition of traffic road signs in an intelligent way, and also demonstrates the performance of Transfer Learning for object classification in general. When systems are trained on the aspects of human visualization (HVS), which helps or generates the same decisions, the construct robust and efficient systems. This allows us to avoid many environmental risks, both for weather conditions, such as cloudy or rainy weather that causes obscured vision of signs, but the main objective is to avoid all road risks that are dangerous to achieve road safety, such as accidents due to non-compliance with traffic rules, both for vehicles and passengers. However, simply collecting road signs in different places does not solve the problem, an intelligent system for classifying road signs is needed to improve the safety of people in its environment. This study proposed a traffic road sign classification system that extracts visual characteristics from a Convolution Neural Network (CNN) classification model. This model aims to assign a class to the image of the road sign through the classifier with the most efficient optimized. Then the evaluation of its effectiveness according to several criteria, using the Confusion Matrix and the classification report, with an in-depth analysis of the results obtained by the images that are taken from the urban world. The results obtained by the system are encouraging in comparison with the systems developed in the scientific literature, for example, the Advanced Driving Assistance Systems (ADAS) of the sector automobile.
Highlights
Machine learning, which is one of the sub-domains of Artificial Intelligence, aims at automatically extracting and exploiting the information present in a dataset for the automobile sector [1]
We explore two different Convolutional Neural Network architectures namely VGG-16, ResNet-34
The observation that it was logical to state that the deeper, the better concerning Convolutional Neural Networks. This makes sense, models should be more capable, their flexibility to adapt to any space increases, a larger space of parameters to explore
Summary
Machine learning, which is one of the sub-domains of Artificial Intelligence, aims at automatically extracting and exploiting the information present in a dataset for the automobile sector [1]. It differs from traditional approaches and facilitates the use of machines or systems in building models from sample data to automate decision-making processes based on the data entered [2]. Training a Convolutional Neural Network is very expensive. That's why hardware manufacturers are stepping up efforts to provide high-performance, GPU graphics processors that can quickly drive a Deep Neural Network by parallelizing the computations [3]. Transfer Learning allows you to do Deep Learning without having to spend a calculating
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
