In the past few years, Deep learning has emerged as an enormous technology which has applications in Image classification and Natural language processing, Recommendation System, Automatic Machine Translation, Handwriting Recognition, etc. Traffic Sign Detection (TSD) systems are used to regulate the drivers during driving and to increase the road safety by reducing accidents. Automatic detection of traffic signs is also important for automated intelligent driving vehicle or driver assistance systems. In previous Traffic Sign Detection Systems, locations of the Traffic Signs are estimated using an object pose elimination problem and the problem is modelled using Convolution Neural Network (CNN). In this work, we are proposing a model using Capsule Network (CapsNet) which overcomes the limitations of CNN and provides training accuracy of 95 percent and training accuracy of 98 percent.
Paper
Full text
Real-Time Traffic Sign Detection using Capsule Network
Semantic Scholar · Computer Science · 2019
Abstract
In the past few years, Deep learning has emerged as an enormous technology which has applications in Image classification and Natural language processing, Recommendation System, Automatic Machine Translation, Handwriting Recognition, etc. Traffic Sign Detection (TSD) systems are used to regulate the drivers during driving and to increase the road safety by reducing accidents. Automatic detection of traffic signs is also important for automated intelligent driving vehicle or driver assistance systems. In previous Traffic Sign Detection Systems, locations of the Traffic Signs are estimated using an object pose elimination problem and the problem is modelled using Convolution Neural Network (CNN). In this work, we are proposing a model using Capsule Network (CapsNet) which overcomes the limitations of CNN and provides training accuracy of 95 percent and training accuracy of 98 percent.