With motto to learn and undertsand working on Convolutional neural netwroks, i have implemented a basic approach to genelaize the current project dataset.
In this project, convolutional neural networks (CNN) based LeNet-5 archictecture are used to classify traffic signs based on German Traffic Sign Dataset.
LeNet 5 Architecture
With Tensorflow framewrok, I have trained and validated a model that generalizes the various traffic signs in the dataset.
The model has achieved validation accuracy of 95.2% and test accuracy of 94.4%.
The model is also tested on random images from internet and all of them are classified correctly.
