FaceShield aims to make you aware of all those times you touch your face without realising it, helping you lose that unhygienic and potentially infectious habit. It uses machine learning algorithms to process the images from your webcam, and sounds an alert whenever your hands get close to your face. It runs in the background, without the need for an internet connection or a learning/training period (support for macOS and Linux coming soon).
Download the programme using the button above, then launch it. The detection starts after the FaceShield icon appears in your system tray/menu bar (this can take a few seconds). Click the FaceShield icon to display the menu where you can start, stop, configure and exit the programme.
Click “Stop” in the FaceShield menu to stop the detection. It can be restarted by subsequently clicking “Start” in that same menu.
Click “Show Video Feed” in the FaceShield menu to display the video currently captured by your webcam. A blue rectangle should be drawn around your face, and green rectangles around your hands. You can change the hand detection sensitivity threshold with the slider appearing above the video feed (a lower value means more hand detections). Please also note that the webcam position and FPS limit, your skin tone and room lighting will have an impact on the accuracy of the detection algorithms.
Click “Show Video Feed” in the FaceShield menu to display the video currently captured by your webcam, and increase the detection sensitivity threshold with the slider appearing above the video feed.
To lower CPU utilisation, you can set a lower FPS limit via the FaceShield menu.
We’re glad you asked :) If you need extra help kicking those bad face-touching habits, what better than Samuel L. Jackson giving you a piece of his mind when you slip. And if that’s not enough, we added an ultimate deterrent - we’ll let you discover that for yourself. You can switch to a different alert type via the FaceShield menu.
FaceShield was built using the handtracking and face_recognition Python libraries. Many thanks to the creators, maintainers and contributors of these projects.
For questions or feedback, you can email us at contact@sandboks.ltd