This project aims to develop a neural network using TensorFlow to classify traffic signs from images, utilizing the German Traffic Sign Recognition Benchmark (GTSRB) dataset.
-
Data Preparation:
- Images are preprocessed using OpenCV for resizing and normalization.
- Data augmentation techniques are applied to enhance the diversity of the training dataset.
-
Model Development:
- A convolutional neural network (CNN) is constructed with TensorFlow.
- The architecture includes convolutional layers for feature extraction, pooling layers for downsampling, and fully connected layers for final classification.
-
Evaluation:
- The model's performance is validated using a separate test dataset.
- Predictions are made on unseen data to evaluate real-world applicability.
-
Documentation:
- Comprehensive documentation of the experimentation process, including hyperparameter tuning and model iterations.
The GTSRB dataset can be accessed via the following links:
This implementation includes several enhancements for improved model performance and additional functionality.
For further information, visit the project page.