Traffic-Sign-Recognition

In this report, we propose a Convolutional Neural Network(CNN) for traffic sign classification that achieves outstanding performance using GTSRB.Since the traffic signs are various in number and design it can be difficult for one to memorize all of them correctly. This development presents a model that identifies a Traffic sign accurately, recognition is carried out in three stages: image preprocessing, feature detection , and recognition. The developed system is specifically designed to detect images invariants in viewing angle, rotation, variable lighting, very low false positive rate and computational time along with GUI speaking the recognised sign name out, which will be really helpful to alert the driver irrespective of his knowledge of traffic signs.

CNN Model

Contributors

• @TirthShah
• @Vidhayini
• @Isha Jain

Deployment

To deploy this project run:

• datasettraffic folder is the GTSRB dataset which is used to train the model;
• Run The traffic_model.ipynb;
• Then used the saved model i.e., my_model.h5 to recognise signs using GUI;
• signs.py is used to store the list of classified signs for reference;
• To run the GUI, run Front-end.py file.

Demo

Link to the video of Front-end: https://youtu.be/iNIGG2xzUZA

Environment Variables

To run this project, you will need to add the following libraries to your venv:-

openCV numpy keras matplotlib sci-kit learn pillow tkinter

Documentation

Paper

Poster