In this project, a system that can accurately detect and determine the type of road sign in a picture is implemented. A convolutional neural network is designed and trained using the GTSRB - German Traffic Sign Recognition Benchmark dataset available on Kaggle with 43 different classes in order to be able to identify the type of road sign in a given image.
- An 97% accuracy was obtained from the final model.
- A simple web application is then developed using the Streamlit library where an image can be uploaded and the corresponding type of road sign displayed.
- This web app was deployed using Heroku and can be found by visiting https://morning-anchorage-96843.herokuapp.com
- The code and results obtained in building and training the model in question can be found in GTSRB_CNN_Results.pdf.
- The code for the web application can be found in the "web_app" directory.
- "Report.pdf" is a comprehensive report describing the technical details of the project.