This project involves creating and comparing the performance of a Convolutional Neural Network (CNN) and a Fully Connected Neural Network (FCNN) on the German Traffic Sign Recognition dataset.
The German Traffic Sign Recognition dataset contains over 50,000 images of 43 different traffic signs. The resolution of the images varies between 15 x 15 pixels and 250 x 250 pixels. All images are provided with an adjusted size of 32 x 32 pixels.
The files train.p and valid.p provide a training dataset and a validation dataset. The training dataset is used to train the model, and the validation dataset is used to validate the model. Once a suitable model has been trained, good results will be obtained on the training and validation data. However, whether the model will achieve good results for new data, i.e., later when the model is used in an application, is unknown. To find out, there are test data. We will test your model at the end on the test data. Therefore, only the training and validation data are provided.
- Clone this repository.
- Install the required packages.
- Run the Python scripts to train the CNN and FCNN models.
- Compare the performance of both models.