🛡️ Intelligent Shoplifting Detection

🖊️ Overview

A cutting-edge deep learning system designed to enhance retail security by detecting and preventing shoplifting in real time. This project leverages YOLO for object detection, DeepSORT for multi-object tracking, and LSTM for activity recognition, ensuring efficient and accurate surveillance.

🚀 Features

• Real-Time Detection: Identifies objects such as people and products instantly.
• Behavioral Analysis: Recognizes suspicious actions like concealing items.
• Anomaly Detection: Flags unusual patterns using autoencoders.
• Multi-Object Tracking: Tracks individuals consistently across frames.

---

🔧 Prerequisites

Before starting, ensure the following software is installed:

1. Python 3.8+
2. pip (Python package manager)
3. CUDA Toolkit (if using GPU acceleration)
4. Supported deep learning frameworks like TensorFlow or PyTorch.

---

🔍 Files to Download

1. Pre-trained YOLO Weights

• Download from the official YOLO website or GitHub: YOLOv8 Pre-trained Weights
• Save the file in the src/models/ directory.

2. Sample Datasets

• Download annotated datasets for training and testing:
  • COCO Dataset
  • Custom dataset (if available): Place in the data/ folder.

3. DeepSORT Model Files

• Download the appearance feature extractor: DeepSORT Weights
  • Save the .pb or .pth file in the src/models/ directory.

4. Pre-trained LSTM Model

• If using pre-trained LSTM for activity recognition:
  • Place the file in the src/models/ directory.

---

🚪 Folder Structure

Shoplifting-Detection/
│
├── README.md          # Project documentation
├── LICENSE            # License information
├── .gitignore         # Ignored files
├── requirements.txt   # Python dependencies
├── data/              # Dataset folder
│   ├── train/         # Training dataset
│   └── test/          # Testing dataset
├── src/               # Source code
│   ├── models/        # Pre-trained models and weights
│   ├── utils/         # Utility functions
│   └── main.py        # Main script
├── notebooks/         # Jupyter notebooks for experiments
├── results/           # Output logs, results, and images
└── docs/              # Documentation files

---

🚜 Installation

Step 1: Clone the Repository

git clone https://github.com/username/Shoplifting-Detection.git
cd Shoplifting-Detection

Step 2: Install Dependencies

pip install -r requirements.txt

Step 3: Add Pre-trained Files

Place the required files (YOLO weights, DeepSORT model, etc.) in their respective folders as described above.

Step 4: Run the System

Execute the main script:

python src/main.py

---

🌐 Usage

• Real-Time Monitoring: Connect to a live surveillance camera and start detection.

  python src/main.py --mode live

• Analyze Recorded Video:

  python src/main.py --mode video --path data/test/video.mp4

---

🔒 License

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

---

🙏 Contributions

We welcome contributions! Follow these steps to contribute:

1. Fork the repository.
2. Create a new branch (feature-name).
3. Submit a pull request.

---

🔎 Acknowledgments

• YOLO for object detection.
• DeepSORT for multi-object tracking.
• COCO Dataset for training data.