danyukezz/plant-disease-classification

🌱 Plant Disease Recognition

An interactive demo for plant disease classification using deep learning.

🎯 About

This app uses a ResNet50 CNN model trained on plant disease images to classify plant health status. It can detect:

• 🟢 Healthy plants
• 🟡 Powdery mildew disease
• 🔴 Rust disease

🏗️ MLOps Pipeline

This model is part of a complete MLOps project featuring:

1. Azure ML Training Pipeline

   • Automated data preprocessing
   • Model training with hyperparameter tuning
   • Model versioning and registry

2. GitHub Actions CI/CD

   • Automatic training on code push
   • Docker containerization
   • Kubernetes deployment

3. Production Deployment

   • FastAPI REST API
   • Kubernetes orchestration
   • Load balancing with 2 replicas

4. Interactive Demo

   • This HuggingFace Space with Gradio interface

🚀 Technology Stack

• Framework: PyTorch
• Model: ResNet50 (pretrained, fine-tuned)
• Training: Azure Machine Learning
• CI/CD: GitHub Actions
• Production: Docker + Kubernetes
• Demo: Gradio on HuggingFace Spaces

📊 Model Details

• Architecture: ResNet50 with custom classifier head
• Input Size: 224x224 RGB images
• Classes: 3 (healthy, powdery, rust)
• Training: Azure ML compute cluster
• Framework: PyTorch 2.0+

🔗 Links

• GitHub Repository: mlops_project_plants
• Production API: Deployed on Kubernetes
• Azure ML Workspace: Model training and management

👨‍💻 Author

Danylo Bordunov
MLOps Project - December 2025

📝 License

MIT License