itsazza/KTT-DAY2-Model-mobilenet_v3_small

🌿 Crop Disease Classifier (MobileNetV3-Small)

This model is a compact, quantized image classifier developed for the AIMS KTT Hackathon. It identifies five common conditions in maize, cassava, and beans, optimized for edge-AI deployment in rural environments.

📊 Model Performance

• Macro-F1 (Clean Test): 0.9933
• Model Size: ~9.2 MB (ONNX)
• Architecture: MobileNetV3-Small (INT8 Optimized)

📂 Classes Supported

The model predicts the following categories:

1. healthy (Maize)
2. maize_rust
3. maize_blight
4. cassava_mosaic
5. bean_spot

🚀 Usage

This model is intended to be served via FastAPI using the onnxruntime library.

---

Note: This model is part of the Tier 2.1 challenge. See process_log.md in the GitHub repo for full reproduction steps.