GreenNest is an innovative cross-platform solution (Mobile & Web) that leverages advanced Deep Learning to help farmers, agriculturists, and plant enthusiasts detect plant leaf diseases instantly with high accuracy. By combining a state-of-the-art AI pipeline with actionable remediation suggestions (powered by Gemini LLM) and a rich plant species library, GreenNest promotes sustainable, data-driven farming practices and supports global food security.
- Full Project Documentation: Google Docs
- API Documentation: Postman
- High Classification Accuracy: Achieved 98.78% accuracy in plant disease identification using ResNet-50 with transfer learning.
-
Three-Stage AI Pipeline:
- YOLOv8: Robust leaf object detection and localization.
- ResNet-50: Fine-grained disease classification.
- Gemini 2.5 LLM: Generates actionable disease remediation suggestions.
-
Cross-Platform Accessibility: Fully functional on Mobile (Flutter) and Web platforms.
-
Real-Time Detection: Immediate diagnostics and treatment plans upon image upload.
-
Comprehensive Species Library: Access a database of 460,000+ plant species with detailed information.
-
Health Monitoring Dashboard: Track historical health and growth of user-defined “Tracked Plants.”
GreenNest is built on a modular architecture with a Node.js MVC core and a dedicated Flask backend for AI.
| Component | Technology | Role |
|---|---|---|
| Backend | Node.js (Express) | Application logic, user authentication, data storage, API endpoints |
| AI Backend | Flask, PyTorch, torchvision, NumPy, Pandas | Serves Deep Learning models (ResNet-50, YOLOv8) |
| Frontend (Web) | EJS | Web user interface |
| Frontend (Mobile) | Flutter | Cross-platform development for Android & iOS |
| AI / LLM | ResNet-50, YOLOv8, Gemini 2.5 LLM | Disease classification, object detection, remediation suggestions |
| Hosting | GCP Compute Engine, Nginx, Cloud Buckets | App hosting, proxy, and storage for user-uploaded images |
| Database | MongoDB | Stores user data, scan history, and plant library |
The system follows a Model-View-Controller (MVC) design for the Node.js backend to ensure scalability, maintainability, and separation of concerns.
High-Level Components:
- Users: Interact with the Load Balancer (via Web/Mobile).
- Load Balancer: Routes requests efficiently.
- Detection Service: Core logic for request validation and coordination.
- Triple model setup: Processes images, performs detection, and returns results.
