NadhemBenhadjali/MIIA-Pothole-Image-Classification-Challenge

Identify which images of roads in South Africa contain potholes # Pothole Image Classification Challenge Notebook ## Introduction This repository contains a Jupyter Notebook for the "MIIA Pothole Image Classification Challenge" hosted on Zindi Africa. The goal of this challenge is to develop a model that accurately classifies images to detect potholes on roads. In this notebook, we use the Fast.ai library built on PyTorch for deep learning tasks. The data has been preprocessed by cropping to adjust and augment it, and a DenseNet-121 architecture has been employed and fine-tuned for classification. ## Dataset The dataset for this challenge is provided by Zindi Africa and consists of images labeled as either containing potholes or not. The data has been preprocessed to ensure uniformity and quality. ## Requirements To run this notebook, the following dependencies are required: - Python 3.x - PyTorch - Fast.ai - Other dependencies as specified in the notebook ## Notebook Contents 1. **Data Preprocessing**: Data is loaded and preprocessed. Cropping and augmentation techniques are applied to improve model performance and generalization. 2. **Model Architecture**: DenseNet-121 architecture is utilized for image classification. 3. **Training**: The model is trained on the preprocessed data. 4. **Evaluation**: Model performance is evaluated using appropriate metrics. 5. **Inference**: Prediction is made on test data. Feel free to customize this README file further with additional details or instructions specific to your notebook or environment.