Part game-of-chance, part animation experience, People Watcher uses your device's camera to map your selected characters and stories onto the locations of the people around you. You can select various modes depending on how people appear through your camera: people walking by, people standing still in a group, a single person, and random.
This app uses the ml5 API to leverage tensorflow.js and poseNet() and the p5 API to work with the HTML canvas element. PoseNet() estimates human poses and outputs the locations for keypoints (hip, shoulder, nose, etc) which can be used to calculate the locations of selected DOM elements (in this case, images and animations).
Project Members: Monica Gonzalez