homo per machina

Welcome! Upload a portrait to see what it looks like to machine learning models, or neural networks, that have no concept of "human." RunwayML provides more information abou the two models used here: SPADE-COCO and SPADE-Landscapes. SPADE-COCO labels objects in images, and SPADE-LANDSCAPES labels landscape elements. RunwayML uses these models to modify photos, telling SPADE where to place and object by drawing a shape with its corresponding color. Misusing these tools by feeding in portraits as if they were those drawn shapes gives a proxy for what the models 'see' in us.