Don't you just love experimenting with code?

hackathon

We do! Especially during hackathons and innovation days. Here are some of our experiments to play around with. Some of them rely on browser flags to be set.

  • Native Browser Face Detection Parallax Art

    avatar

    Dave Bitter

    Google Developer Expert Web & Developer Advocate

    This application uses the experimental Face Detection API in the browser to bring artwork to life. It takes the position of the face and uses it to create a parallax effect on the top layer of the artwork.

    The experimental Face Detection API is part of the browser. There is no additional code shipped to make it work. This opens up a powerful toolset for developers to leverage and create concepts around their applications.

    Read more about Native Browser Face Detection.

  • Snakeface.js

    avatar

    Ye Jun Wu

    Senior Frontend Consultant

    This application was built in a few hours during one of our Google Days where we get to try out different technologies and hack concepts together. It was built using ML5.js to bring Machine Learning (ML) to the browser through an abstracted layer. This results in this demo fully running on the client with ease of use for the developer due to the abstraction layer. Next to that, it uses the library P5.js to make it a breeze to build a game around it.

    This demo makes use of P5.js and ML5.js which go hand in hand and makes building interactive concepts with ML and gamification very approachable for developers.

  • TensorFlow.js Gym Pose with Squad Counter

    avatar

    Dave Bitter

    Google Developer Expert Web & Developer Advocate

    This application was built in a few hours during one of our Google Days where we get to try out different technologies and hack concepts together. TensorFlow.js brings Machine Learning (ML) to the browser. This results in this demo fully running on the client of the user without the need for a server to execute the TensorFlow logic.

    This demo uses WebGL as the “back-end” for TensorFlow. A trained ML model on the human body and pose is then passed to retrieve the data points from. After that, you can both visualise and build concepts around this data.