The goal of this project is to develop an AI-powered face spoofing detection system using YOLO-based real-time face detection and computer vision techniques. The system is designed to differentiate between real and fake faces, detecting whether an individual in front of the camera is authentic or a spoof (e.g., a photograph, video, or mask). The project aims to provide a robust solution for face recognition security by identifying potential spoofing attempts in real-time.
- Real-time Face Detection 🕵️♂️: Detects faces in video streams using YOLOv5 for accurate and fast results.
- Spoof Detection 🚫📸: Differentiates between real faces and spoofed attempts (e.g., photos or videos) to prevent unauthorized access.
- Confidence Scoring 💯: Each detection is accompanied by a confidence score that helps assess the accuracy of the detection.
- Bounding Box Visualization 📦: Draws bounding boxes around detected faces, making it easier to see which faces are being recognized.
- Customizable Detection Threshold ⚙️: Adjustable confidence threshold to filter out low-confidence detections.
├── SpoofShield
| ├── Code/
│ ├── data.py
│ ├── main.py
│ ├── split_data.py
│ └──train.py
│ ├── Datasets/
│ │ ├── All
│ │ ├── DataCollection
│ │ ├── Real
│ │ ├── Fake
│ │ └── SplitData
│ ├── Test/
│ │ ├── face_detector_test.py
│ │ ├── test.txt
│ │ ├── text_file_test.py
│ │ └── yolo_test.py
│ ├── models/
│ └──l_version_1_300.pt
├── .gitignore
├── LICENSE
└── README.md
Watch the project demo here:
github.mp4
- YOLOv5 for real-time face detection
- OpenCV for image processing and computer vision tasks
- Python for scripting and automation
- cvzone for simplifying OpenCV tasks like bounding box drawing