YOLO-Based Object Detection with FastAPI - Final Year Project

This project is a YOLO (You Only Look Once) object detection system built using FastAPI. It provides real-time predictions and visualizations, leveraging the YOLO model for detecting specific activities or objects. This project is developed for a final-year academic project.

Features

• Object Detection: Predict objects in uploaded images using a custom YOLO model (best5.pt).
• Custom Trained Classes: Detects the following classes:
  • Drinking
  • Eating
  • Violence
  • Sleeping
  • Smoking
  • Walking
  • Weapon
• Real-Time Prediction API: Upload images and get annotated predictions with bounding boxes and confidence scores.
• Database Storage: Stores prediction results (class name, accuracy, and timestamp) in a database.
• Data Visualization: View class-wise prediction counts in a bar chart.

Technology Stack

• Backend Framework: FastAPI
• Model: YOLO (Ultralytics)
• Database: SQLAlchemy with SQLite
• Frontend: Jinja2 Templates, HTML, CSS, JavaScript (Bootstrap, Chart.js)

How to Run the Project

Prerequisites

1. Python 3.8+ installed.
2. Install dependencies using the requirements.txt file.

Steps to Run

1. Clone the repository:

   git clone https://github.com/alihassanml/Yolo-Final-Year-Project.git
   cd Yolo-Final-Year-Project

2. Create and activate a virtual environment:

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

3. Install dependencies:

   pip install -r requirements.txt

4. Start the FastAPI server:

   uvicorn app:app --reload

5. Open your browser and navigate to:

   http://127.0.0.1:8000
   

6. Use the web interface to upload images and view predictions.

API Endpoints

• GET /
  Returns the homepage.

• POST /predict
  Upload an image and receive predictions with annotated bounding boxes.

• GET /predictions
  Retrieve a list of all predictions stored in the database.

• GET /class-counts
  Get class-wise counts of predictions for visualization.

Project Structure

Yolo-Final-Year-Project/
│
├── app.py                 # Main FastAPI application
├── model/
│   ├── best5.pt           # YOLO pre-trained model weights
│   └── __init__.py        # Package initialization
│
├── database.py            # Database configuration
├── model.py               # SQLAlchemy models
├── static/
│   ├── style.css          # CSS for styling
│   ├── script.js          # JavaScript functionality
│   └── uploads/           # Uploaded images folder
│
├── templates/
│   └── index.html         # Jinja2 template for the frontend
│
├── requirements.txt       # Python dependencies
├── README.md              # Project documentation
└── .gitignore             # Files to ignore in the repository

Future Enhancements

• Add video-based detection for live streams.
• Integrate a user authentication system for secure access.
• Expand the model to detect additional classes.
• Include export functionality for predictions (e.g., CSV or Excel).

License

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

---

Author

Developed by Ali Hassan. For questions or feedback, feel free to reach out.

---

### `requirements.txt`

```plaintext
fastapi==0.95.2
uvicorn==0.23.0
sqlalchemy==2.0.21
jinja2==3.1.2
pydantic==2.0.3
opencv-python==4.8.0.74
numpy==1.24.3
ultralytics==8.0.30
Pillow==9.5.0
chart.js==3.9.1

---

Steps to Upload to GitHub

1. Save the README.md and requirements.txt in the root directory of your project.
2. Commit and push the changes to your GitHub repository:

   git add README.md requirements.txt
   git commit -m "Added README and requirements.txt for FastAPI-based YOLO project"
   git push origin main