No materials for the event yet, sorry!

In this workshop, we'll share a curriculum around AI that we're currently working on. We try to show how to use machine learning in the classroom by implementing a gesture recognizer (based on the $1 gesture recognizer) in Snap!.

We start by creating a single-stroke gesture drawing program. By building an "animate" control structure based on the pen trails, we are able to animate our drawn gestures. To recognize the drawn gestures, we extend the drawing program to record paths. We then discuss why it's difficult to just compare two paths to recognize a gesture, and introduce the "resample" block that normalizes all paths to a specific number of points. The actual recognition happens in a nearest neighbour block that will find the smallest difference between the current resampled path and the paths stored in an examples list. Finally, we can use this identified gesture for storytelling by broadcasting the name of the gesture and animating it accordingly.

We will also share our pedagogical approaches, lesson plans, and discussion ideas as well as further AI education resources in this workshop.