The Maskade Classifier is a lightweight computer vision project built with C++. It uses the user's camera to determine if they have a face mask for COVID-19 correctly placed on their face. It also features a minigame in which the user has to try to keep their face behind a floating digital mask to score points before time runs out. The name is a pun on cascade classifiers (e.g. the Haar Cascade Classifier), a common type of computer vision object detection model.

• Plain classification: run the program and it will tell you if you're wearing a mask or not! Avoid dark lighting for best results.
• Minigame: Press the M key to enter or exit the minigame mode, where you can take off your real mask and score points by keeping your face behind a floating mask for as many frames as possible within the time limit. Press the R key to reset the game score and timer while in progress.
• JSON configuration file that allows you to set default parameters for the program's style and fonts. This can be found in config/config.json.
• The program uses a pretrained TensorFlow savedmodel file to classify images from your computer's camera feed. This can be found ing assets/converted_savedmodel. This model was created using the Google Teachable Machines project.

The program can be cloned and run locally, but it requires several libraries to be installed and linked on your computer to properly work. The dependencies are listed below:

The above libraries can be installed using brew install or via direct download. Once the appropriate libraries have been installed and built, the CMakeLists.txt file can be used to link them. Simply replace the corresponding file paths with your local paths to the libraries.

Finally, the CMakeLists.txt file can be used to build the project to a build folder of your choosing a C++ compiler of your choosing. The maskade-visualizer target should be run afterwards. This can either be accomplished through the command line or using an IDE or text editor with support for C++ and CMake, such as CLion or Visual Studio Code.

This project was developed on MacOS 11.1 with the clang 12.0.0 compiler, but it can be expected to work on Windows or Linux, or with a different compiler such as gcc. An MIT License is included. It was created during my time at the University of Illinois at Urbana-Champaign for my CS 126: Software Design Studio class.

Please reach me at rustom.dev or [email protected] with any questions or thoughts. Thank you!