KrishiMitra AI — The Farmer's Copilot 🇮🇳

A unified, data-driven assistant for Indian farmers, designed to provide actionable, real-time advice on the most critical aspects of agriculture.

This project brings together a suite of powerful tools to answer a farmer's most pressing questions, from market prices to crop health, all through a simple, accessible interface.

UPDATE:-

Demo of our 20+ indian local language Voice Assisted chat feature

---

✨ Core Features

• 📈 Market Price Forecasting: Fetches live prices from Agmarknet (data.gov.in) and uses ML to provide a farmer-friendly Sell / Wait recommendation for the next 1-2 weeks.
• 🛰️ Real-Time Satellite Analysis: Leverages Sentinel-2 data to analyze vegetation health (NDVI, NDMI, NDWI, LAI), automatically adjusting the area of interest to find cloud-free images.
• 🌦️ Hyperlocal Weather Forecasts: Provides detailed 7-day and 24-hour weather summaries for the farmer's specific location.
• 📚 Fact-Grounded RAG: Uses a Retrieval-Augmented Generation pipeline over a curated Agri Knowledge Base (seeded with official government documents and agricultural university guidelines) to answer complex queries without hallucination.
• 🌍 Location Intelligence:
  • geocode.py: Converts location names (e.g., "Kharagpur") into precise latitude and longitude using OpenStreetMap.
  • geo.py: Performs reverse geocoding to identify the state, district, etc., from geographic coordinates.
• 🗣️ Multilingual Support:
  • lang.py: Automatically detects the user's language and provides answers in the same language for a natural, intuitive experience.
• 🌿 Leaf Disease Detection: A local Vision Transformer (ViT) model identifies common plant diseases from a photo, with the LLM providing tailored care advice.
• 🤖 Accessible Bots: Deployed via a Telegram bot with a user-friendly interface, including voice commands (STT/TTS) and quick-action buttons.

---

🚀 Quick Start

Clone the Repository

git clone <your-repo-url> && cd krishimitra

Set Up Python Environment

python -m venv venv
# On Windows
venv\Scripts\activate
# On macOS/Linux
source venv/bin/activate

Install Dependencies

pip install -r requirements.txt

Create and Configure .env File

# On Windows
copy .env.example .env
# On macOS/Linux
cp .env.example .env

Now, open the .env file and fill in your secret API keys.

Build the RAG Index

python backend/app/rag/index.py --rebuild

Run the Backend API

uvicorn backend.app.main:app --reload

The API will be available at http://127.0.0.1:8000.

Run the Telegram Bot (in a new terminal)

python bots/telegram/bot.py

---

🏗️ What’s Inside (Architecture)

backend/
  app/
    services/pipeline.py    # Orchestrates tools + RAG → final answer
    tools/
      mandi.py              # data.gov.in (Agmarknet) client + caching
      pricing.py            # Quantile LightGBM (p20/p50/p80) + SELL/WAIT
      weather.py            # 24h + 7d forecast summary
      sentinel.py           # Sentinel Hub indices (NDVI/NDMI/NDWI/LAI)
      lang.py               # Language detection and translation
      geocode.py            # Location name → Lat/Lon
      geo.py                # Lat/Lon → State/District
    rag/
      index.py              # Chroma index build/load
      retrieve.py           # Top-k passages
      generate.py           # LLM synthesis (tool+RAG aware)
    utils/cache.py          # Simple in-memory TTL cache
bots/
  telegram/bot.py           # Chat, slot filling, voice, disease photos
vit_disease/
  vit_model.py              # Local ViT classifier (leaf disease)
  llm_helper.py             # Care advice text via LLM
models/
  pricing_global/           # p20/p50/p80 models + meta + encoder

---

💻 Tech Stack

🔑 Environment Variables (.env)

Use .env.example as a template. You will need to provide the following keys:

# --- OpenAI (RAG + generation) ---
OPENAI_API_KEY=sk-...
OPENAI_EMBED_MODEL=text-embedding-3-small
OPENAI_CHAT_MODEL=gpt-4o-mini

# --- Agmarknet (data.gov.in) ---
DATA_GOV_IN_API_KEY=your_api_key

# --- Sentinel Hub ---
SH_CLIENT_ID=your_client_id
SH_CLIENT_SECRET=your_client_secret

# --- Telegram ---
TELEGRAM_BOT_TOKEN=123456:ABC...

# --- Backend URL (for bot -> backend) ---
BACKEND_URL=http://127.0.0.1:8000

---

🧪 How to Test Locally

Backend API

Health Check:

curl -s http://127.0.0.1:8000/health

Full-Featured Query:

curl -s -X POST http://127.0.0.1:8000/ask \
  -H 'content-type: application/json' \
  -d '{
    "text": "Should I sell tomatoes now?",
    "crop": "Tomato",
    "state": "Karnataka",
    "district": "Bangalore",
    "market": "Ramanagara",
    "geo": {"lat": 12.522, "lon": 76.897},
    "horizon_days": 7,
    "debug": true
  }' | jq .

🌊 Architecture Flow Diagram

Telegram Bot

Start a chat with your bot and try these commands:

• Share your location (📍)

• Ask a question:

• "कल बारिश होगी?"

• "What seed variety suits this unpredictable weather?"

• "क्या मुझे इस हफ्ते टमाटर बेचना चाहिए?"

• "Will next week’s temperature drop kill my yield?"

• "Kal ka mausam kaisa rahega?"

• Send a photo of a diseased plant leaf.

• Send a voice note with your question.

---

📸 Key Features in Action

Here are some screenshots showcasing the KrishiMitra AI bot in action.

• Welcome & Location. Onboarding new users with multilingual examples.