Plant-Disease-Detection-Management-System

KO | EN

gitlite — search

#react #typescript #go #linux #nodejs #java #windows #kubernetes #vue #pytorch #security #ios #mcp #open-source

Plant-Disease-Detection-Management-System

(★ 20)

DEMETER is an AI-powered plant disease detection web app that helps farmers identify crop diseases early through image analysis. Users register, enter basic crop details, and upload leaf images. A trained deep learning model analyzes the photo and predicts disease presence, offering actionable recommendations.

Plant-Disease-Detection-Management-System 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

#  DEMETER — Plant Disease Detection & Management System

DEMETER is an AI-powered web platform that helps farmers and growers identify plant diseases from a leaf photo, understand the severity of infection, and get treatment guidance — with a community and marketplace layer built around it.

> Named after the Greek goddess of agriculture, DEMETER combines deep learning (CNN-based image classification) with classical computer vision (HSV-based infection region detection) to give both a **diagnosis** and a **visual severity map** from a single leaf image.

---

##  Features

| Feature | Status | Description |
|---|---|---|
| 🔍 **Disease Detection** | ✅ Implemented | Upload a leaf photo → CNN classifies the disease (trained on the PlantVillage dataset) |
| 🎯 **Infection Region Mapping** | ✅ Implemented | OpenCV/HSV-based lesion detection highlights affected areas and estimates infection % |
| 📊 **Prediction History** | ✅ Implemented | Every scan is logged with timestamp, crop, disease, and confidence — viewable, exportable (CSV/JSON), and deletable |
| 🔐 **User Accounts** | 🟡 Partial | Login/registration flow exists; session-based access control is still being hardened |
| 👥 **Farmer Community** | 🟡 Prototype (UI only) | Discussion boards and farming tips UI — not yet backend-connected |
| 🛒 **Marketplace** | 🟡 Prototype (UI only) | Product browsing and cart UI (client-side only) — no backend order system yet |
| 👤 **Farmer Profile** | 🟡 Prototype (UI only) | Profile editing UI, not yet tied to the account system |
| 📱 **Mobile-ready model export** | 🧪 Experimental | TensorFlow.js model export explored for on-device/browser inference |

**Legend:** ✅ Fully working · 🟡 Frontend built, backend integration pending ·  Exploratory

---

##  How It Works

```
   Leaf Photo
       │
       ▼
┌─────────────────┐        ┌──────────────────────┐
│  CNN Classifier  │──────▶│  Disease + Confidence │
│ (TensorFlow/Keras)│       └──────────────────────┘
└─────────────────┘
       │
       ▼
┌─────────────────┐        ┌──────────────────────┐
│  OpenCV / HSV    │──────▶│  Infection % + Overlay│
│  Segmentation    │       │  (highlighted regions)│
└─────────────────┘        └──────────────────────┘
       │
       ▼
┌─────────────────┐
│  Disease Info DB │──▶  Symptoms, cause, and treatment recommendations
└─────────────────┘
```

1. **Classification** — A CNN trained on the PlantVillage dataset predicts the disease class from the uploaded image.
2. **Segmentation** — An HSV color-thresholding pipeline (OpenCV) detects lesion-like regions (browning, yellowing, powdery patches, necrotic spots), draws contours around them, and computes an infection percentage.
3. **Guidance** — The predicted disease is matched against a disease information database for symptoms and treatment suggestions.
4. **Logging** — The result is saved to prediction history for tracking over time.

---

##  Tech Stack

| Layer | Technology |
|---|---|
| Backend | Flask (Python) |
| ML / Classification | TensorFlow / Keras (CNN) |
| Computer Vision | OpenCV |
| Frontend | HTML, CSS, JavaScript (server-rendered templates) |
| Data | JSON-based storage (disease info, prediction history) |
| Experimental | TensorFlow.js (client-side inference exploration) |

---

##  Project Structure

```
DEMETER_APP/
├── app.py                    # Flask application & routes
├── class_indices.json        # Class label mapping
├── disease_info.json         # Disease metadata (symptoms, treatment)
├── plant_disease_model.keras # Trained CNN model
├── plant_disease_model.h5    # Trained CNN model (H5 format)
├── templates/                # Frontend pages
│   ├── login.html
│   ├── register.html
│   ├── dashboard.html
│   ├── disease.html          # Detection & upload page
│   ├── history.html          # Prediction history
│   ├── community.html        # Farmer community (UI prototype)
│   ├── market.html            # Marketplace (UI prototype)
│   └── profile.html           # User profile (UI prototype)
├── static/                   # CSS, JS, images
├── model/                    # TF.js model export (experimental)
└── tensorflowjs-model/       # TF.js model export (experimental)
```

---

##  Getting Started

### Prerequisites
- Python 3.9+
- pip

### Installation

```bash
# Clone the repository
git clone https://github.com/HarrickChristoJP/Plant-Disease-Detection-Management-System.git
cd Plant-Disease-Detection-Management-System

# Install dependencies
pip install -r requirements.txt
```

> **Note:** If `requirements.txt` isn't present yet, install the core dependencies manually:
> ```bash
> pip install flask tensorflow opencv-python numpy pillow
> ```

### Running the App

```bash
python app.py
```

The app will be available at `http://127.0.0.1:5000`.

### Verifying the Model

A model verification script (`verify_model.py`) is included to confirm the model's architecture, class mapping, and real prediction confidence:

```bash
python verify_model.py --image path/to/leaf.jpg
```

---

##  Model Details

- **Dataset:** [PlantVillage](https://www.kaggle.com/datasets/emmarex/plantdisease) — labeled leaf images across multiple crops and disease classes
- **Architecture:** Convolutional Neural Network (CNN) built with TensorFlow/Keras
- **Input:** RGB leaf images
- **Output:** Disease class + confidence score

> ⚠️ **Model validation in progress.** Formal accuracy/F1 benchmarks on a held-out test set are being finalized and will be published here once complete.

---

##  Roadmap

- [ ] Publish model accuracy benchmarks (confusion matrix, per-class F1)
- [ ] Wire Community and Marketplace pages to a real backend and database
- [ ] Add per-user session enforcement across all protected routes
- [ ] Expand disease coverage beyond current crop set
- [ ] Migrate prediction history from JSON storage to a proper database
- [ ] Add automated tests and CI

---

##  Known Limitations

- Community, Marketplace, and Profile pages are currently frontend prototypes without full backend integration
- Prediction history storage is file-based and not yet optimized for concurrent multi-user access
- Model accuracy has not yet been formally benchmarked and published

---

##  License

This project is currently unlicensed. If you intend to reuse, modify, or build on this code, please reach out first.

---

##  Author

**Harrick Christo JP**
B.Tech, Artificial Intelligence & Data Science
St. Joseph's College of Engineering, Chennai

---

##  Contributing

This is currently a solo student project under active development. Suggestions and issue reports are welcome via GitHub Issues.

Plant-Disease-Detection-Management-System

Plant-Disease-Detection-Management-System 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Plant-Disease-Detection-Management-System

Plant-Disease-Detection-Management-System 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README