Traffic Sign Image Classification

Overview

The Traffic Sign Image Classification project is a machine learning application designed to classify traffic signs using a Convolutional Neural Network (CNN). Developed as a Jupyter Notebook, this project leverages TensorFlow and Keras to build and train a CNN model that accurately predicts the class of traffic signs from images. The model is trained on the "GTSRB - German Traffic Sign Recognition Benchmark" dataset and achieves an accuracy of approximately 94% on the test set.

Features

• Convolutional Neural Network: Utilizes a CNN model built with TensorFlow and Keras for image classification.
• High Accuracy: Achieves ~94% accuracy on the test dataset, demonstrating effective learning and prediction capabilities.
• Comprehensive Dataset: Trained on 39,209 images spanning 43 classes and evaluated on 12,630 test images.
• Statistical Evaluation: Employs various statistical methods to evaluate model performance, including accuracy score, confusion matrix, and classification reports.
• Research-Based: Built upon extensive research, including insights from 45 studies on CNNs, image classification, and data preprocessing.

Dataset

The project uses the GTSRB - German Traffic Sign Recognition Benchmark dataset from Kaggle, which includes:

• Training Set: 39,209 images
• Test Set: 12,630 images
• Classes: 43 different traffic sign categories

Technology Stack

• Python: Programming language for development.
• TensorFlow & Keras: Libraries for building and training the CNN model.
• Jupyter Notebook: Interactive environment for code development and testing.
• scikit-learn: Used for statistical analysis and evaluation.
• seaborn: Library for visualizing the confusion matrix and other data plots.