Crop Disease Detection & Smart Irrigation Hub

An intelligent, autonomous agronomy system that analyzes crop leaves for infections and manages farm water valves dynamically using customizable rules. Designed for deployment on local edge controllers (e.g., Raspberry Pi 5) and desktop environments.

Developed with ❤️ and dedicated to premium visual layout aesthetics.

---

🏗️ System Architecture

This project implements a Secure Backend-Proxy Inference model. To safeguard developer credentials, the private Roboflow API Key is kept entirely in the backend .env variables and is never exposed to the client-side JavaScript.

Data Flow Pipeline

flowchart TD
    A[Webcam / Drag-Drop Image] -->|1. Base64 POST| B[Flask API Endpoint]
    B -->|2. Reads Secret Key| C{Environment / DB}
    C -->|API Key Present| D[Roboflow Inference SDK Client]
    C -->|Offline Mode| E[Local Simulated Prediction]
    D -->|3. Serverless Inference| F[Roboflow API Server]
    F -->|4. Predictions JSON| D
    D -->|5. Bounding Box Coordinates| G[Inference Router]
    E -->|5. Mock Coordinates| G
    G -->|6. SQL Insert| H[(SQLite Database)]
    G -->|7. Evaluate Rules| I[Smart Rules Matcher]
    I -->|8. Toggle Relay| J[GPIO Valve Controller]
    I -->|9. Warning Alert| K[SMTP Email Alerts]
    G -->|10. Final JSON Payload| L[Browser Client JS]
    L -->|11. Draw Boxes| M[HTML5 Canvas Overlay]
    L -->|12. Speak Warning| N[Text-to-Speech Output]

Loading

---

🌟 Key Features

1. 🔒 Secure Backend Inference: All model invocations proxy through the local Python environment. The Roboflow API keys are masked inside administrative pages.
2. 🤖 Smart Rules Engine: Custom rules engine allowing admins to match specific crop species and disease states against confidence thresholds to execute actions:
   • Start Irrigation: Cycles valves for $X$ seconds.
   • Stop Irrigation: Turn off valves (essential to prevent fungal spores from multiplying on wet foliage).
   • Notify Only: Dispatch asynchronous SMTP email alerts to the agronomist.
   • Do Nothing: Log metrics for passive observation.
3. 🌦️ Weather Microclimate risk advisor: Directly queries the free Open-Meteo weather API using the client's browser geolocated coordinates to calculate crop humidity-risk warnings.
4. 📊 Analytics Dashboards: Built-in Apache ECharts displaying timeline distributions, disease severity percentages (🟢 Normal, 🟡 Moderate, 🔴 Severe), and irrigation log histories.
5. 📁 Batch Image Processor: Drag-and-drop file interface featuring sequential uploads, asynchronous queue logs, progress trackers, and aggregated summary reports.
6. 📜 Pagination Timeline: Interactive diagnosis history panel supporting search filters, original-vs-detected compare slider handles, PDF diagnostic report exporters, and CSV raw data downloads.

---

🧪 Researched Pesticide Dosage Rates

The hub incorporates precise chemical treatment advisory thresholds to assist farmers in remediation:

Crop Class	Diagnosed Disease	Recommended Pesticide	Precise Dosage Rate	Frequency
Tomato / Potato	Late Blight	Mancozeb (Dithane M-45)	2.0 - 2.5 grams / Liter	Every 5-7 days in wet weather
Tomato / Potato	Early Blight	Chlorothalonil (Daconil)	1.5 - 2.0 mL / Liter	Every 7-10 days at first spot
Corn	Common Rust	Pyraclostrobin (Headline)	0.8 - 1.2 mL / Liter	Apply at pustule; repeat in 14d
Apple	Apple Rust	Myclobutanil (Immunox)	1.25 mL / Liter	Petal fall, weekly for 3 weeks
Apple	Apple Scab	Captan 80 WDG / Sulfur	2.5 grams / Liter	Green tip, weekly for 2 cycles
All Crops	Generic Rust	Copper Octanoate Soap	1.5 grams / Liter	Spray foliage every 7-10 days
All Crops	Generic Scab	Lime-Sulfur spray	2.0 mL / Liter	Apply weekly during damp spring

---

🛠️ Local Setup and Run

1. Requirements

• Python 3.9+ (Fully compatible up to Python 3.14)
• SQLite3

2. Installation

Clone the project repository, initialize a Python virtual environment, and install dependencies:

# Create and activate virtual environment
python3 -m venv .venv
source .venv/bin/activate

# Install dependencies (incorporates inference-sdk and visual requirements)
pip install -r requirements.txt

3. Environment Configuration

Create a .env file in the root folder to securely configure environment variables:

ROBOFLOW_API_KEY=your_private_api_key

4. Running the App

Start the Flask web app server:

python3 app.py

Open http://localhost:5001/ in your browser.

• Admin Role: Username admin / Password admin
• Farmer Role: Username farmer / Password farmer

---

⚙️ Hardware Controls (Raspberry Pi 5)

When run on a Raspberry Pi 5, the app attempts to interact with GPIO Pin 18 using gpiozero to control the physical irrigation relay. On non-Pi environments (such as Mac, Linux, or Windows PCs), it gracefully fails over to a Mock GPIO Driver, printing valve state updates to the terminal console so that full product presentations run flawlessly.

---

The Final Output

---

Made By Kuldeep Sisodiya❤️