Project Archived

EagleEye has been archived. The development of a web interface and major updates have moved to a new repository. Please visit ArguX for the latest version and ongoing improvements.

EagleEye

Real-Time Object Detection in Public Camera Feeds

EagleEye is a Python script that scans public cameras gathered from Insecam to locate objects of interest via YOLOv4 object detection model.

---

Features

• Multithreaded for scanning multiple streams simultaneously.
• Provides the detected camera's IP address, geolocation, and the model's confidence level in identifying the specified object.
• Filter searches by country.
• Option to open detected camera streams directly in your web browser.

---

Installation

1. Clone the repository:

   git clone https://github.com/erfangolpour/EagleEye.git

2. Install the required dependencies:

   pip install -r requirements.txt

3. Download the YOLOv4 weights file (245 MB):

   • yolov4.weights
   • Google Drive mirror: yolov4.weights

   Place the file in the YOLO directory.

---

Usage

Run the script with the following options:

usage: EagleEye.py [-h] [-n NUMBER] -o {person,bicycle,car,...} [-w WORKERS] [-c COUNTRY]

EagleEye - A tool for scanning public security cameras

Options:
  -h, --help            Show this help message and exit.
  -n NUMBER, --number NUMBER
                        Maximum number of cameras to scan (default: 100).
  -o {person,bicycle,car,...}
                        Specify the object to search for (refer to YOLOv4 - Darknet/coco.names for options).
  -w WORKERS, --workers WORKERS
                        Limit the number of multiprocessing pool workers.
  -c COUNTRY, --country COUNTRY
                        Filter results by country.

You can find the list of searchable objects here.

---

Demonstration

The object of interest here is set to car:

---

Configuration

The following settings can be customized in the script:

• MIN_CONFIDENCE: Minimum confidence threshold for object detection (default: 0.6).
• OPTIMIZED_MAX_WORKERS: Default number of worker threads (CPU cores minus one).
• REQUEST_TIMEOUT: Timeout duration for web requests in seconds (default: 10).
• USER_AGENT: User agent string for web requests.

---

Contributing

Contributions are welcome! If you encounter any issues or have ideas for improvement, feel free to open an issue or submit a pull request.

---

License

This project is licensed under the GPLv3 License.

---

Acknowledgments

Special thanks to:

• The YOLO object detection model.
• The Insecam website for public camera streams.
• The tqdm library for progress bar visualization.