🌿 AgriNova — AI-Powered Smart Agriculture Ecosystem
Smart Farming, Better Harvest
A complete AgriTech system combining drone-based crop health monitoring, real-time AI detection, and farmer alerts — built for Indian farmers.
🚀 What is AgriNova?
AgriNova is a full-stack smart agriculture ecosystem where a drone flies over a farm, detects crop diseases in real time using AI, maps every tree's health status, and instantly alerts the farmer via WhatsApp — all from a mobile app.
🏗️ System Architecture
Drone Camera (PiCamera2)
↓
Raspberry Pi 4 ←→ YOLOv8 Model (Edge AI)
↓ ↓
Pixhawk GPS Disease Detection
↓ ↓
Flask Server (Port 8000) on Pi
↓
Android Mobile App
├── Live Drone Feed (MJPEG Stream)
├── Real-Time Farm Grid (Farm Tree Map)
├── Gallery Scan (Plantix-style)
├── AI Expert Diagnosis
├── Agri Market
└── Farm Area Calculator (GPS)
↓
WhatsApp Alerts via Twilio
├── Per-detection alert (pest + confidence + GPS)
└── Full farm scan summary report
✨ Features
🚁 Drone Module
- Live Drone Feed — Real-time MJPEG video stream from PiCamera2 to mobile app
- Edge AI Detection — YOLOv8 runs directly on Raspberry Pi (no cloud needed)
- Drone Capture — Capture image on demand with instant AI analysis
- GPS Geotagging — Every detection tagged with GPS coordinates via Pixhawk flight controller
- Detection Logging — All detections saved as images + JSON with GPS + Google Maps link
🗺️ Farm Grid
- Live Custom Tree Health Map — Each tree shown as Affected / Healthy / Pending input is taken from farm
- Configurable Grid — Set rows, columns, frames per tree, and detection threshold at runtime
- Auto-refreshes every 5 seconds — Live updates as drone scans the farm
- Visual Web Dashboard — Flask serves a live HTML grid page
📱 Mobile App (Android)
- Multi-language — English, Marathi, Hindi
- Home Dashboard — Live weather, your crops, daily farming tips
- Drone Control Panel — Drone Capture, Live Stream, Gallery Scan, Live Farm Grid
- Gallery Scan — Farmer uploads any crop photo → instant pest/disease detection
- AI Expert — Describe problem + upload photo → AI gives diagnosis and advice
- Agri Market — Buy seeds, pesticides, fertilizers directly in app (₹)
- Farm Area Calculator — Calculate field area via GPS in Acres, Hectares, Guntha
- User Authentication — Login/Register system
💬 WhatsApp Alerts (Twilio)
- Real-time pest alert — Sent when disease detected with name, confidence %, GPS, timestamp
- Full scan summary — Sent after complete farm scan with total affected/healthy trees, disease breakdown, and Google Maps location link
🦠 Diseases Detected
| # | Disease | Crop |
|---|---|---|
| 1 | Powdery Mildew | Grapes, Tomato |
| 2 | Downy Mildew | Grapes |
| 3 | Early Blight | Tomato |
🔄 Coming Soon: Nutritional deficiency detection (dataset ready, model training in progress)
🛠️ Tech Stack
AI & Computer Vision
- YOLOv8 — Object detection model (trained on custom dataset)
- OpenCV — Image processing and video streaming
- Python — Core backend language
Hardware
- Raspberry Pi 4 — Edge computing, runs detection + Flask server
- PiCamera2 — Drone-mounted camera
- Pixhawk Flight Controller — GPS via MAVLink protocol
Backend & API
- Flask — REST API + MJPEG video streaming server on Pi
- PyMAVLink — Drone GPS communication
- Twilio — WhatsApp alert integration
Mobile App
- Android — Native mobile application
- HTTP — Connects to Flask server on Raspberry Pi
Model Training
- Google Colab — YOLOv8 model training
- Custom Dataset — Annotated crop disease images
📡 Flask API Endpoints (Runs on Raspberry Pi)
| Endpoint | Description |
|---|---|
/video_feed |
Live MJPEG drone stream |
/latest_detection |
Latest detected frame as JPEG |
/detection_info |
Latest pest detection as JSON |
/status |
System status + GPS info |
/log |
Full geotagged detection log |
/tree_health |
Farm scan summary JSON |
/farm_grid |
Full grid status JSON |
/grid_view |
🌿 Visual farm health webpage |
⚙️ Setup & Configuration
Requirements
pip install ultralytics picamera2 pymavlink twilio flask opencv-python
Configuration
Before running, set your credentials in agrinova_modified.py:
TWILIO_SID = "your_twilio_sid_here"
TWILIO_TOKEN = "your_twilio_token_here"
TWILIO_FROM = "whatsapp:+XXXXXXXXXXXXX"
FARMER_NUMBER = "whatsapp:+91XXXXXXXXXX"
Run on Raspberry Pi
python agrinova_modified.py
You will be prompted to enter:
- Number of rows and columns (farm grid size)
- Frames per tree (how many frames to analyze per tree)
- Detection threshold % (e.g. 20 means 20% pest frames = affected)
🏆 Achievements
- 🥇 1st Place — Open Innovation Track, Sambhaji Rao Jondhale College of Engineering, Dombivli
- 🏆 Winner — ExpoTech Competition, Vision 2026, Walchand College of Engineering, Sangli
- 🥈 Runner-Up — Ideathon, Vision 2026, Walchand College of Engineering, Sangli
👨💻 Team
Built by a team of 4 — Robotics & Automation Engineering students at Walchand College of Engineering, Sangli
📄 Files in This Repository
| File | Description |
|---|---|
agrinova_modified.py |
Main Raspberry Pi script — detection + Flask server + WhatsApp alerts |
galleryscanfeature.py |
Gallery scan feature — farmer uploads photo for detection |
collabmodeltraining.ipynb |
Google Colab notebook for YOLOv8 model training |
changingdefaultlabels.ipynb |
Notebook for customizing dataset labels |
data12.yaml |
YOLOv8 dataset configuration file |
⚠️
best.pt(trained model) is not included in this repo due to file size. Contact us to get the model.
AgriNova v1.0 • Made for Indian Farmers 🌾
Built With
- jupyter-notebook
- python
