promibe/Dataverse-Africa-Cassava-Mosaic-Detector

Dataverse Africa Cassava Mosaic Detection system # Dataverse-Africa-Cassava-Mosaic-Detector Dataverse Africa Cassava Mosaic Detection system # Cassava Mosaic Disease Detection System This is a Proof-of-Concept (POC) for Dataverse Africa — a deep learning-based image classification system that detects Cassava Mosaic Disease using transfer learning with MobileNetV2. ## 🚀 Project Overview Cassava is a vital crop in Africa, and early detection of diseases like Cassava Mosaic can greatly enhance food security. This system classifies cassava leaf images into: 1. **Cassava Mosaic** 🦠 2. **Healthy Cassava** 🌿 3. **Non-Cassava Plant** 🍀 The solution uses TensorFlow and MobileNetV2 for transfer learning, achieving over **98% accuracy** on the validation set. --- ## 📁 Project Structure ``` ├── image.png # Background image used ├── cassava disease detection system.py # Main Streamlit app ├── runtime # python version used ├── data_split/ # Structured training/validation/test data ├── cassava_classifier_final.h5 # Best model saved as .h5 ├── requirements.txt # Dependency list ├── .gitignore # Files to ignore in version control └── README.md # This file ``` --- ## 🛠 Features - **Image Classification** using transfer learning (MobileNetV2) - **Interactive Predictions** via Streamlit UI - **Custom Image Upload Support** - **Real-Time Accuracy & F1 Score Reporting** - **Confusion Matrix & Visualization** --- ## ⚙️ Installation ```bash git clone github.com cd dataverse-africa-cassava-mosaic-detector pip install -r requirements.txt streamlit run "cassava disease detection system.py" ``` > Make sure to use **Python 3.10 or 3.11**, as TensorFlow does not yet support Python 3.13. --- ## 📷 Sample Predictions You can test your own images by placing them in the `test_plants/` folder or using …