AI-Powered Fertilizer Deficiency Detection is a full-stack intelligent agriculture system that enables farmers and agronomists to upload plant/soil images or input crop data and instantly detect nutrient deficiencies — then receive AI-driven fertilizer recommendations — without needing any technical expertise.
Identifying fertilizer deficiencies manually requires expert agronomists, lab testing, and days of waiting. This platform compresses that entire process — from raw data to actionable fertilizer recommendations — into seconds using computer vision and AI.
🔹 Core Detection Engine
- Deep learning-based plant disease & nutrient deficiency detection
- Multi-class classification (Nitrogen, Phosphorus, Potassium, Micronutrient deficiencies)
- Supports both image uploads and structured soil/crop data input
🔹 AI Intelligence Layer
- Computer Vision Model — detect deficiencies directly from leaf/plant images
- Soil Data Analyzer — quality scoring based on NPK levels and pH values
- AI Chat Assistant — ask questions about your crop health in natural language
- Fertilizer Recommendation Engine — tailored recommendations based on crop type and deficiency
- Auto PDF Reports — one-click professional diagnosis report generation
🔹 Production & Web Features
- Full-Stack Web App — Next.js frontend with FastAPI/Python backend
- REST API — prediction endpoint for external integrations
- Role-Based Access — farmer, agronomist, and admin roles
- Crop History Tracking — maintain diagnosis history per field/crop cycle
- Responsive UI — works on desktop and mobile devices
- Dockerized Deployment — production-ready with Docker Compose
🏠 Dashboard
|
📷 Image Upload & Detection.jpeg)
|
🧪 Soil Data Analysis.jpeg)
|
🌿 Deficiency Results
|
💊 Fertilizer Recommendation Panel
|
|
| Layer | Technology |
|---|---|
| Backend | FastAPI (Python) |
| Frontend | Next.js + TypeScript |
| Database | MySQL 8.0 |
| ML / CV Libraries | TensorFlow / PyTorch, OpenCV, Scikit-learn |
| Image Processing | Pillow, OpenCV |
| Containerization | Docker + Docker Compose |
| Reverse Proxy | Nginx |
| Auth | JWT + RBAC |
| Reports | ReportLab (PDF) |
# 1. Clone the repo
git clone https://github.com/MuhammadAdnan586/Fertilizer_Deficiency_Detection.git
cd Fertilizer_Deficiency_Detection
# 2. Setup environment
cp .env.example .env
# Edit .env with your DB password, SECRET_KEY, ANTHROPIC_API_KEY
# 3. Deploy with Docker
docker-compose up --build| Service | URL |
|---|---|
| Frontend App | http://localhost |
| API Docs (Swagger) | http://localhost:8000/docs |
| Detection API | http://localhost:8000/api/detect |
Fertilizer_Deficiency_Detection/
├── FertilizerProject/
│ ├── backend/
│ │ ├── api/ # FastAPI routes & endpoints
│ │ ├── models/ # ML model definitions & weights
│ │ ├── services/ # Detection & recommendation logic
│ │ └── main.py # Application entry point
│ ├── frontend/
│ │ ├── components/ # React UI components
│ │ ├── pages/ # Next.js pages
│ │ └── styles/ # CSS / Tailwind styles
│ ├── ml/
│ │ ├── training/ # Model training scripts
│ │ ├── datasets/ # Dataset preparation
│ │ └── evaluate.py # Model evaluation
│ └── docker-compose.yml
├── .gitignore
└── README.md
| Nutrient | Deficiency Symptoms Detected |
|---|---|
| Nitrogen (N) | Yellowing of older leaves, stunted growth |
| Phosphorus (P) | Purple/reddish discoloration, poor root development |
| Potassium (K) | Leaf edge scorching, weak stems |
| Magnesium (Mg) | Interveinal chlorosis on older leaves |
| Iron (Fe) | Interveinal chlorosis on young leaves |
| Calcium (Ca) | Tip burn, blossom-end rot |
import requests
# Detect deficiency from image
with open("leaf_image.jpg", "rb") as f:
response = requests.post(
"http://localhost:8000/api/detect",
files={"image": f},
data={"crop_type": "wheat"}
)
result = response.json()
print(result["deficiency"]) # e.g. "Nitrogen Deficiency"
print(result["confidence"]) # e.g. 0.94
print(result["recommendation"]) # Fertilizer recommendation| Model | Accuracy | Precision | Recall | F1-Score |
|---|---|---|---|---|
| CNN (Custom) | 91.2% | 90.8% | 91.5% | 91.1% |
| ResNet-50 | 94.7% | 94.2% | 95.1% | 94.6% |
| EfficientNet-B3 | 96.3% | 96.0% | 96.5% | 96.2% |
Contributions are welcome! Please feel free to submit a Pull Request.
- Fork the project
- Create your feature branch (
git checkout -b feature/AmazingFeature) - Commit your changes (
git commit -m 'Add some AmazingFeature') - Push to the branch (
git push origin feature/AmazingFeature) - Open a Pull Request
Distributed under the MIT License. See LICENSE for more information.
Muhammad Adnan
