Semaphore is an open-source computer vision project on GitHub that turns your whole body into a keyboard: by holding your arms in flag semaphore positions in front of a webcam, you can type letters, digits and symbols hands-free. The project has around 1,900 stars and comes from the Everything Is Hacked maker channel. Note this is the gesture-typing experiment, not the CI or automation tools that share the Semaphore name.
The software uses OpenCV together with Google’s MediaPipe pose detection to track body landmarks in real time from ordinary video input. It calculates the angles of both arms, matches them against a dictionary of flag semaphore positions, and converts each recognized pose into the corresponding keystroke sent to the computer. Rather than forcing users to hold every pose for a fixed delay, it includes a jump shortcut to repeat the last symbol, which makes sustained typing more practical. The repository includes a demo video and well-documented code mapping arm angles to characters, making it an approachable reference project for pose-estimation input methods.
Semaphore is completely free and open source under the GPL-3.0 license. Anyone can download it from GitHub and run it locally; there are no accounts, paid tiers or hosted components involved.
Semaphore appeals to makers, accessibility experimenters and computer vision students interested in alternative input methods. It is a fun, educational way to explore pose estimation, and a starting point for building gesture-controlled interfaces of your own.