Recommendations
TouchDesigner, Beautiful Soup, ml5.js, Socket.IO
2019

Recommendations is an interactive installation that automatically matches viewers with a product on the Amazon marketplace – referencing (in a somewhat tongue-in-cheek way) the seemingly limitless capability of modern e-commerce to suggest, advertise, and match consumers with products they supposedly “need”.

Recommendations was built using TouchDesigner, Beautiful Soup (Python library), ml5.js (p5.js library) and Socket.IO. The installation uses a webcam connected to the ml5.js library via a locally hosted p5.js sketch to “recognize” objects in front of the camera. This information is then sent via Socket.IO to an instance of Beautiful Soup within TouchDesigner, which uses the object names as keywords to search the marketplace. The results are then displayed to the viewer onscreen.

© Jack DiLaura