Highly sensitive Deep Learning Model for Road Traffic Sign Identification
DOI:
https://doi.org/10.17762/msea.v71i4.875Abstract
Computer vision and artificial intelligence plays major role in avoiding road accidents and large death numbers due to attention lack of driver. Recognition of traffic signs may help drivers to get alert. In this project recognition of traffic signs such as ‘Go’, ‘Straight’, ‘Speed Limit’, ‘stop’, ‘No passing’, etc. are used. We used a deep learning method known CNN with the help of LeNet architecture for the traffic sign classification and identification. Google Colaboratorywhich offers cloud based Jupyter Notebook environment that allows us to use supporting packages like NumPy, TensorFlow, keras, GPU etc. We have optedabove design to absorb thefeatures of the dataset such as contour based or liveliness-basedqualities by considering features such asscalability, rotation, luminous changes.By feeding GTSRB dataset to it Accuracy of the designed model is achieved to 98.50 %.