🌾 KrishiMitra AI

Multilingual AI-powered agricultural intelligence system delivered through WhatsApp

KrishiMitra is an AI-assisted agricultural intelligence platform designed to help farmers with:

• crop disease diagnosis
• mandi price assistance
• weather intelligence
• multilingual conversational support

through a WhatsApp-based workflow powered by FastAPI, TensorFlow, Gemini Vision, and Twilio APIs.

---

🔗 Live Demo & Deployments

🌐 Frontend

https://krishi-mitra-dev-ai.vercel.app/

🤗 Backend Deployment

https://shivamr021-krishimitra-ai.hf.space/

🎥 Demo Video

https://youtube.com/shorts/ygmrwU6daT0?feature=share

---

✨ Core Features

🐛 Crop Disease Diagnosis

Implements a confidence-threshold-based hybrid inference workflow:

1. A local TensorFlow/EfficientNet model performs fast disease classification.
2. Low-confidence predictions are routed to Gemini Vision for additional analysis.

This architecture improves robustness while reducing unnecessary API calls.

---

📈 Real-Time Mandi Price Assistance

Users can request agricultural market pricing using natural-language WhatsApp queries.

Example

"What is the soybean price in Indore?"

---

🌤 Weather Intelligence

Provides real-time weather insights including:

• temperature
• humidity
• environmental conditions

to support day-to-day agricultural planning.

---

🌐 Multilingual AI Interaction

Supports multilingual conversational responses including:

• Hindi
• English

using Gemini-powered conversational workflows.

---

⚙️ Technology Stack

Component	Technology
Backend API	FastAPI
ML Framework	TensorFlow / Keras
Vision Model	EfficientNetB0
LLM Integration	Google Gemini
Messaging Platform	Twilio WhatsApp API
Deployment	Hugging Face Spaces
Frontend Hosting	Vercel
Containerization	Docker

---

🏗 System Architecture

KrishiMitra routes user requests through specialized processing pipelines:

• Disease Detection Pipeline
• Market Price Handler
• Weather Intelligence Handler
• Conversational AI Handler

The disease diagnosis workflow combines:

• local CNN inference
• confidence-threshold routing
• Gemini Vision fallback analysis

to improve handling of uncertain or low-quality agricultural images.

Detailed system design notes are available in:

docs/ARCHITECTURE.md

---

📊 Model Evaluation

The disease classification pipeline was trained using transfer learning on PlantDoc-style agricultural datasets.

Evaluation artifacts and observations are available in:

docs/EVALUATION.md
docs/EVALUATION_RESULTS.md
evaluation/eval.py

Key Observations

• controlled datasets generalized poorly to real-world farmer images
• compressed WhatsApp images remain challenging
• low-light agricultural images reduced prediction reliability
• fallback routing improved uncertain prediction handling

---

🖼 Screenshots

Frontend Demo

WhatsApp Workflow

Hindi Interaction

---

🚀 Local Setup

Clone Repository

git clone https://github.com/shivamr021/KrishiMitra-AI.git

cd KrishiMitra-AI

---

Install Dependencies

pip install -r requirements.txt

---

Configure Environment Variables

Create a .env file:

GEMINI_API_KEY=
TWILIO_ACCOUNT_SID=
TWILIO_AUTH_TOKEN=
WEATHER_API_KEY=

---

Run Backend Server

uvicorn main:app --reload

---

💬 Example Interaction

User

मक्का में कौनसी बीमारी है?

Bot

आपकी फसल में फंगल संक्रमण के संकेत दिखाई दे रहे हैं। कृपया कॉपर-आधारित फफूंदनाशी का उपयोग करें।

---

📁 Repository Structure

backend/
frontend/
docs/
evaluation/
assets/

---

👥 Team

Shivam Rathod

Backend Development, AI Integration, TensorFlow Model Training, FastAPI Development, Twilio Workflow Automation

• LinkedIn: https://www.linkedin.com/in/shivamrathod021

• GitHub: https://github.com/shivamr021

---

Shatakshi Tiwari

Backend Support, AI/ML Integration Support, Project Ideation, Research & Presentation Design

• LinkedIn: https://www.linkedin.com/in/shatakshitiwari017/

• GitHub: https://github.com/Shatakshi0216

---

Sahil Kukreja

Frontend Development, UI Integration, Weather API Configuration Support

• LinkedIn: https://www.linkedin.com/in/sahil-kukreja-943993289/

• GitHub: https://github.com/Sahilkukreja30

---

Nitika Jain

Documentation, Presentation Support, and Research Assistance

• LinkedIn: https://www.linkedin.com/in/nitika-jain-b8690b353/

• GitHub: https://github.com/nitikajain25

---

📌 Project Context

This project was initially developed during the OpenAI × NxtWave Buildathon and later refined into an applied AI systems engineering portfolio project.

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.