Image Segmentation with Road Traffic Sign Det...

Image Segmentation with Road Traffic Sign Detection Using Convolutional Neural Networks and Deep Learning

Author Name : Pathoori Nheha Chaturvedi, K Yasudha, Dr. Vanitha Kakollu

ABSTRACT

This project is a vision-based vehicle guidance system. Traffic signs play a very crucial role in autonomous driving vehicles. These signs provide drivers with essential information about the road, in order to make driving safer and easier. This project has two main phases. The first phase is the detection, uses colour thresholding to segment the image and uses shape analysis to detect the signs. The second phase is for the classification, uses a convolutional neural network. Most of the work is done by focusing on recognizing signs leaving important information provided by other type of signs like text-based signs. The detection phase uses Image Processing method that creates contours on each video frame and finds all ellipses and circles among those contours. They are marked as candidates for traffic signs and then images are created by cropping from the original frame based on candidates coordinate in the classification phase. A pre-trained SVM model is used to classify these images to find out which type of traffic sign the image contains.

Keywords: Convolutional Neural Network, Autonomous driving vehicle, Support Vector Machine.