🌾 PRAGATI PATH — AI-Powered Agricultural Education Platform

PRAGATI PATH is a full-stack web platform designed to assist Indian farmers by leveraging AI technology, multilingual support, and image recognition. It provides crop suggestions, disease diagnosis, and educational content to improve agricultural outcomes.

💡 Key Features

• 🤖 AI Chatbot using Google Gemini 2.5 Flash for farming-related query resolution in local languages
• 🌱 Crop Recommendation System based on user location and weather data
• 🗺️ "My Farm" Interactive Map — Leaflet.js pin-drop map where farmers mark their exact plot. GPS coordinates are saved to MongoDB and automatically passed to the entire agronomy pipeline (SoilGrids, Open-Meteo, Mandi prices) — no manual input needed.
• 🦠 Plant Disease Detection (Dual Mode)
  • Primary: Gemini Vision (Analyzes any crop, provides treatment/prevention)
  • Fallback: Offline MobileNetV3-Large (128 unified classes) running natively in-browser via LiteRT WebGPU (Fast, private, works offline)
• 🌍 Deep Agronomy Engine — Integrates SoilGrids (soil chemistry) and Open-Meteo (5-year historical climate) for hyper-local precision agriculture. Results cached in localStorage for 24 hours to eliminate redundant API calls.
• 📊 Market Integration — Real-time APMC Mandi commodity prices via data.gov.in.
  • Zero-Cost Synergy: Utilizes OpenWeather's Reverse Geocoding API to dynamically translate raw GPS coordinates into the strict text-based state/district format required by the Indian Government API, seamlessly bridging two incompatible systems.
• 💬 Krishi Charcha Forum — A community discussion board for farmers to share tips, crop alerts, and ask questions.
• 🏛️ Government Schemes Tracker — Static database of 7 national schemes (PM-KISAN, PMFBY, KCC, PMKSY, eNAM, Soil Health Card, PM-KUSUM) with eligibility info, helplines, and direct apply links
  • ⚠️ Deadline Banners — Auto-shows PMFBY Kharif/Rabi or PM-KISAN installment alerts based on current month (pure date math, zero API)
  • 🤖 AI Eligibility Check — Gemini ranks all 7 schemes by priority for the farmer's state, crops, and land ownership
• 🗓️ Smart Farming Calendar — Daily learning tracker with a Crop Timeline Overlay: pinned crops show their sowing/growing/harvest phase for each calendar month (zero API calls)
• 🌡️ Weather-Aware Greeting — Contextual farming tip (e.g. "🌧️ Rain today — delay pesticides") shown after weather loads, using already-fetched data; zero extra API requests
• 📤 WhatsApp Share — One-tap share of plant disease scan results (works on both Gemini and offline AI modes)
• 🌐 Language-Aware YouTube Search — Reads the active Google Translate language and appends it to every YouTube query (supports 9 Indian languages: Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Kannada, Malayalam, Punjabi)
• 📺 YouTube Learning Module — Fetches top relevant videos and tracks course completion progress
• 🎓 Certificate Generator — Completing a course (≥95% watched) unlocks a downloadable PDF certificate of completion via html2pdf.js
• 🔐 Secure login and protected routes using Clerk authentication
• 🔑 BYOK Architecture (Bring Your Own Key) — Gemini API keys are securely provided by users and stored only in their local browser storage, ensuring free hosting and no central API cost.

🌐 Live Preview

🔗 https://pragatipath.onrender.com/

🛠 Tech Stack

• Frontend: HTML, Vanilla CSS, JavaScript, Leaflet.js (interactive map)
• Backend: Bun, Express.js
• Database: MongoDB (via Mongoose)
• AI Integration: Google Gemini (Generative AI), LiteRT.js (WebGPU WASM)
• Auth: Clerk
• PDF Export: html2pdf.js (Certificate Generator)

🏗 Architecture

flowchart TB
    subgraph Client ["Frontend Browser"]
        UI["Dashboard UI (Vanilla JS)"]
        Map["Leaflet.js (My Farm Map)"]
        Storage["LocalStorage (BYOK Key + 24hr Agronomy Cache)"]
        OfflineAI["LiteRT WebGPU (MobileNetV3)"]
    end

    subgraph Backend ["Bun + Express Server"]
        API["API Proxy (aiSearch.js)"]
        AuthGuard["Clerk Middleware"]
        DBHandler["Mongoose (DBHandler.js)"]
    end

    subgraph External ["External Services"]
        ClerkAuth["Clerk Identity"]
        Gemini["Google Gemini 2.5"]
        MongoDB[("MongoDB Atlas")]
        WeatherAPI["OpenWeather Geo API"]
        GovData["data.gov.in (APMC)"]
        SoilGrids["SoilGrids API"]
        OpenMeteo["Open-Meteo"]
        YouTubeAPI["YouTube Data V3"]
    end

    %% Frontend to Backend
    Map -->|"Save farm lat/lon"| API
    UI -->|"lat/lon coordinates"| API
    Storage -..->|"Injects X-Gemini-Key Header"| API
    
    %% Offline Fallback
    UI -->|"Image Upload (No Key)"| OfflineAI

    %% Auth & DB Flow
    API --> AuthGuard
    AuthGuard -->|"Verify Token"| ClerkAuth
    API -->|"User Progress / Forum"| DBHandler
    DBHandler <--> MongoDB

    %% The Reverse Geocoding Bridge (Genius Workaround)
    API -->|"1. Reverse Geocode lat/lon"| WeatherAPI
    WeatherAPI -->|"2. Returns State/District"| API
    API -->|"3. Query using State String"| GovData

    %% Agronomy Intelligence Pipeline
    API -->|"Fetch pH/Nitrogen"| SoilGrids
    API -->|"Fetch 5-yr Climate"| OpenMeteo
    API -->|"Bundled Soil+Weather JSON"| Gemini

    %% Schemes (static, no external API)
    UI -->|"GET /api/schemes"| API
    UI -->|"POST /api/schemes/eligibility"| Gemini

    %% Video (language-aware)
    API -->|"Localised Search Queries"| YouTubeAPI

Loading

📦 Getting Started

Prerequisites

• Bun (Runtime and package manager)

Installation

git clone https://github.com/vikashgupta16/PragatiPath.git
cd PragatiPath
bun install

Environment Variables

Create a .env file in the root directory:

CLERK_SECRET_KEY=your_secret_key
CLERK_PUBLISHABLE_KEY=your_publishable_key
CLERK_SIGN_IN_URL=/public/Accounts/signin.html
CLERK_SIGN_UP_URL=/public/Accounts/signup.html
MONGO_URI=your_mongodb_connection_string
YOUTUBE_API_KEY=your_youtube_api_key
OPENWEATHER_API_KEY=your_openweather_api_key
GOV_DATA_API_KEY=your_gov_api_key

Run the App

bun run dev

The app will be running at http://localhost:8080.

📖 Documentation

• How the Offline Plant AI Works

📂 Project Structure

/client               # Frontend (Public Landing, Private Dashboard)
/server               # Express API routes, Auth middleware, Database schemas
/docs                 # Technical documentation
/scripts              # AI Model kaggle scripts, dataset dictionary generators

👨‍💻 Authors

• Rouvik Maji – Backend Developer
• Archis – Backend/Frontend Developer
• Vikash Gupta – Frontend Developer
• Rajbeer Saha – Frontend Developer

📄 License

GPL-3.0 © 2025 Archisman Pal