Original article. Unity Web demo.

This is an adaptation of the original C# code for Unity.

In the demo, only one point-cloud template is loaded for each of the 16 gesture types. You can add additional templates as you wish, and even define your own custom gesture templates.

The $P Point-Cloud Recognizer is a 2-D gesture recognizer designed for rapid prototyping of gesture-based user interfaces. In machine learning terms, $P is an instance-based nearest-neighbor classifier with a Euclidean scoring function, i.e., a geometric template matcher. $P is the latest in the dollar family of recognizers that includes $1 for unistrokes and $N for multistrokes. Although about half of $P's code is from $1, unlike both $1 and $N, $P does not represent gestures as ordered series of points (i.e., strokes), but as unordered point-clouds. By representing gestures as point-clouds, $P can handle both unistrokes and multistrokes equivalently and without the combinatoric overhead of $N. When comparing two point-clouds, $P solves the classic assignment problem between two bipartite graphs using an approximation of the Hungarian algorithm. The $P recognizer is distributed under the New BSD License agreement.

BSD-3