CryptograFace is an innovative C++ application that combines the robust security of facial biometrics with the convenience of data encryption and management. Designed with a user-friendly graphical interface, this project offers a seamless authentication process through facial recognition, ensuring secure access to encrypted text and image files.
Upon logging in, users gain access to their personalized directory, where they can securely browse and manage their encrypted data. Whether safeguarding sensitive documents or protecting personal drawings, CryptograFace provides a powerful solution for users seeking both advanced security measures and intuitive file management capabilities.
CryptograFace incorporates a robust facial recognition system leveraging advanced computer vision techniques. Using OpenCV, the application captures images of users' faces on dedicated worker threads, maintaining responsiveness on the main GUI window. Facial descriptors are then extracted using Dlib's powerful deeplearning neural-network models, generating unique vector representations of facial features. These descriptors are grouped via the Chinese Whispers Clustering algorithm, ensuring the system's ability to identify and authenticate users accurately.
With security at the forefront of considerations, the threshold for facial descriptor matches is fine-tuned to reduce false positives. During user login, the system employs reinforced learning techniques to continuously update the dataset of facial features. This adaptive approach ensures that the system adapts to changes in user appearance and becomes increasingly reliable over time.
In addition to managing text-based data, CryptograFace also offers a unique hand-tracked virtual drawing canvas feature. Leveraging Python's extensive machine learning libraries, including MediaPipe and PyTorch's pretrained models, this capability allows users to interact with a digital canvas directly within a live camera feed. A Python interpreter embedded within the C++ code facilitates real-time hand tracking, enabling precise detection of user finger gestures and positions.
Cryptograface enables users to paint directly on a digital canvas overlaid on the live camera feed. Users can select from multiple colors and brush sizes to create intricate digital artworks. Moreover, painting progress is automatically saved across sessions, with each image securely encrypted to ensure data integrity and privacy.
CryptograFace employs Crypto++'s comprehensive suite of encryption functionalities to secure both text-based and image data using the Advanced Encryption Standard (AES). During encryption, CryptograFace dynamically generates new pairs of private keys and initialization vectors for each operation. This approach enhances security by preventing potential vulnerabilities associated with compromised keys. Moreover, the decryption process in CryptograFace includes checksum verification to detect and prevent corruption of encrypted ciphers. This additional layer of security guarantees the authenticity of decrypted data, providing users with confidence in the integrity of their stored information.
For example,
Lorem ipsum dolor sit amet
might get encoded into
2E8F0EE35012A35784A55A565C637889CE6FB17E3A07AC8506D7206E0ADAEDFBBF0D5F169E6F935257799F5196
CryptograFace features an intuitive graphical user interface (GUI) built with C++'s Qt framework, Embracing a minimalist design philosophy that ensures ease of navigation. The architecture of the GUI is optimised with multi-threading capabilities, ensuring responsiveness even during resource-intensive tasks. Static windows are defined using Qt Modeling Language (QML), while dynamic components such as file directories are generated dynamically at runtime. The GUI not only handles file management operations such as creation, deletion, and renaming, but it also integrates user-friendly features such as pop-up reminders for unsaved content.
Ensure that you have the following installed
Clone this repo with the command
git clone https://github.com/alfaloo/CryptograFace.git
Run the CMakeLists.txt file.
You are now ready to go!
This project is under the MIT license.
Special thanks to the creators of OpenCV, Dlibs, Mediapipe, Crpyto++, and Qt for their amazing libraries/frameworks, without which, the completion of this project would not be possible.
Additionally, I would like to shoutout:
- @packetcode, for their face-recognition-opencv project, from which I have taken great inspiration.
- @MohamedAlaouiMhamdi, for his AI_virtual_Painter project, from which I have taken great inspiration.
- @serengil, for his Fine Tuning The Threshold in Face Recognition analysis, from which I have taken great inspiration.