🌾 CropAI – AI-powered Agricultural Assistant

A Deep Learning + Machine Learning-based system for crop disease detection, crop recommendation, and fertilizer recommendation.

---

🚀 Project Overview

CropAI is a fully integrated, end-to-end AI system for farmers that can:

1. Detect crop disease from an image
2. Recommend the best crop based on soil & climate parameters
3. Recommend the right fertilizer based on crop, soil type, nutrients

Includes:

• ML + DL Models
• FastAPI Backend
• Streamlit Dashboard
• Cloud Deployment
• (Mobile app coming soon)

---

🏷️ Tech Stack Badges

---

🔥 Features

🩻 1. Crop Disease Detection (Computer Vision)

• Trained MobileNetV2 on the PlantVillage dataset
• Input: leaf image
• Output: disease label + confidence

API Endpoint: POST /predict-disease Example Response:

{
  "predicted_class_index": 14,
  "disease_label": "Tomato___Early_blight",
  "confidence": 0.83
}

---

🌱 2. Crop Recommendation System

Uses soil & climate parameters:

• N, P, K
• Temperature
• Humidity
• pH
• Rainfall

API Endpoint: POST /recommend-crop Example Response:

{
  "recommended_crop": "rice"
}

---

🌾 3. Fertilizer Recommendation System

Based on:

• Temperature, Humidity, Moisture
• Soil Type, Crop Type
• N, P, K

API Endpoint: POST /recommend-fertilizer Example Response:

{
  "recommended_fertilizer": "Urea"
}

---

🧱 Project Architecture

graph TD
    A["PlantVillage Image Dataset"] --> B["Crop Disease Model (MobileNetV2)"]
    C["Crop & Fertilizer Datasets"] --> D["Tabular ML Models"]

    B --> E["Saved Models Folder"]
    D --> E

    E --> F["FastAPI Backend"]

    F --> G["Streamlit Dashboard"]
    F --> H["Mobile App (Future)"]
    F --> I["Cloud Deployment (Render/AWS/Railway)"]

Loading

---

🌍 Live API (Render Deployment)

The backend is publicly deployed via Render.

• Base URL: https://cropai-project.onrender.com/
• Swagger Docs: https://cropai-project.onrender.com/docs

⚠️ Note: Render free tier sleeps when idle. First request may take 20–40 seconds.

---

📡 API Reference (Quick)

GET /health

Check server status.

Response:

{ "status": "ok" }

---

POST /predict-disease

Predict disease from uploaded image.

Body: multipart form-data → file

Response example:

{
  "predicted_class_index": 14,
  "disease_label": "Tomato___Early_blight",
  "confidence": 0.83
}

---

POST /recommend-crop

Body:

{
  "N": 90,
  "P": 42,
  "K": 43,
  "temperature": 22.0,
  "humidity": 80.0,
  "ph": 6.5,
  "rainfall": 200.0
}

Response:

{ "recommended_crop": "rice" }

---

POST /recommend-fertilizer

Body:

{
  "Temparature": 26.0,
  "Humidity": 52.0,
  "Moisture": 38.0,
  "Soil_Type": "Loamy",
  "Crop_Type": "Sugarcane",
  "Nitrogen": 50.0,
  "Potassium": 40.0,
  "Phosphorous": 40.0
}

Response:

{ "recommended_fertilizer": "Urea" }

---

🖼️ Screenshots

All screenshots are stored in the /screenshots folder on GitHub.

Dashboard Homepage

Disease Prediction

Crop Recommendation

Fertilizer Recommendation

---

🧠 How to Run Locally

1️⃣ Create environment

conda create -n cropai python=3.10
conda activate cropai
pip install -r requirements.txt

2️⃣ Start the backend

cd backend
uvicorn main:app --reload --host 0.0.0.0 --port 8000

3️⃣ Run Streamlit dashboard

cd dashboard
streamlit run app.py

---

🎯 Future Improvements

• Better disease classification accuracy
• More crops & fertilizers
• Full mobile app (Flutter)
• Multi-language farmer UI
• Deployment across scalable cloud infra
• Improve accuracy with heavy fine-tuning
• Deploy backend + dashboard to Render/Railway/AWS
• Add real-time weather API integration
• Build farmer login + recommendation history

---

📄 License

This project is licensed under the MIT License. Feel free to use it, modify it, and build upon it with proper attribution.

---

💼 Why This Project Matters

This project demonstrates:

• 🧠 Deep Learning (CV with MobileNetV2)
• 📊 Machine Learning (RandomForest, preprocessing)
• ⚙️ API engineering (FastAPI, Pydantic, CORS)
• 🖥️ Interactive dashboards (Streamlit)
• 🧩 Full end-to-end design (data → model → API → UI)
• 🧱 Production-ready architecture

Perfect for:

• Resume
• LinkedIn posts
• Interviews
• Final-year projects
• ML engineering portfolios

---

👤 Author

Akhil Saurabh