Tibor Udvari

Cheerio AI

Your personal AI cheerleader

Cheerio AI is a playful fitness companion capable of learning various exercises. It monitors repetitions and provides ongoing encouragement to boost motivation during workout sessions.


Many people still do not have a clear understanding of the basics of Machine Learning. While tools like Google’s Teachable Machine can offer some intuition, they are not yet commonly used by non-technical individuals in their daily lives.

We created Cheerio AI to demonstrate how simple models can be used in everyday life. We developed an exercise counter, an ideal use case for a classifier. By integrating the training process into the app’s flow, users can quickly train the classifier for their specific one-off applications on the go.

The user trains a classifier to recognize an active and rest state. The application then uses this to count reps and offer encouragement.

1. Train Active State
2. Train Rest State
3. Exercice


The project resulted from a collaboration with Amaury Hamon during a week-long workshop in February 2023. The workshop, “Teachable Machine,” focused on exploring creative applications of machine learning and was led by Matteo Loglio from oio studio. The workshop’s primary objective was to highlight the potential of artificial intelligence for creative pursuits in art and design.

The project uses the ml5js KNNClassifier.


Live Project cheerioai.fun
Source Control Github Repository


Thank you to Andrea Becze-Deák and Benjamin Boulesteix for their help in the video.