This project is an image recognition model that identifies German road signs using a convolutional neural network built and trained with PyTorch. After playing around with some hyperparameters, the final validation accuracy was around 80%.
Here is a demo of the project being used on a video feed of a drive-around in Berlin.
