http://visibleobjectsinvisiblethings.webflow.io/
Computer vision recognizes objects, humans see things.
Ten photographs from the TIME100 most influential photos of all time are passed through two object-detection computer vision models trained on COCO dataset. The objects identified by both models are added up and cropped away. The result is a photograph with two separate layers: a layer of objects that the computer vision models detected and a layer of things that they did not.
![](../wp-content/uploads/2020/11/Screen-Shot-2020-11-18-at-6.44.54-PM-1621x1080.png)
Tools used: Runway ML, Webflow, Photoshop
Personal websites: www.sheungyiu.com; www.pekkovasantola.com