Skip to main content

Raspberry Pi Unlocks Computer by Detecting Push-Ups with ML

Raspberry Pi
(Image credit: Victor Sonck)

If you’re tired of typing in a password to log into your computer and you don't use a fingerprint reader or an IR camera, you can at least get a workout in. Maker Victor Sonck has created a Raspberry Pi-powered push-up authentication project so that you break a sweat when you log in. Instead of logging in with something typical like a string of characters, Sonck logs in with a string of reps using a little help from machine learning (ML) on our favorite single-board computer.

Sonck shared the creation process behind this project through his ML Maker channel on YouTube which at the moment only features this project. However, a quick look at his recent GitHub activity shows a history of ML-based projects leading up to this Pi-powered, exercise-inducing creation.

The Raspberry Pi push-up detection system runs independently from his PC and is positioned in a far corner of the room. Using a camera, it detects when Sonck has successfully completed the number of pushups necessary to log in to his machine before sending a command to allow access.

The project is built around a Raspberry Pi 4 which is capable of processing machine learning applications on its own but to avoid adding to its workload, Sonck opted to use an Oak 1 AI module. This device features a 4K camera alongside an Intel Myriad X chip which can handle additional AI Processing needs for the project. According to Sonck, it connects and interfaces easily with the Pi making it an ideal component for his project needs. The setup also includes a display, microphone and speaker for audio output.

The ML push-up detection system relies on an open-source application called Blazepose which can recognize human body poses from images and builds a skeleton with points marking joint locations to duplicate said poses in real-time. These skeletons are more simple than raw images to interpret which eases the burden on the push-up detection program. The source code is available at GitHub for anyone interested in digging deeper into how it works.

If you want to recreate this Raspberry Pi project and feel the burn for yourself, check out the original video shared to YouTube by Victor Sonck and be sure to follow him for more interesting ML projects.

Ash Hill
Ash Hill

Ash Hill is a Freelance News and Features Writer at Tom's Hardware US. She manages the Pi projects of the month and much of our daily Raspberry Pi reporting.