Is that you? MegaPixels uses a custom built facial recognition system to search for your face amongst the millions of images used for training facial recognition algorithms.
Over the last 15 years researchers have amassed hundreds of datasets by scraping the Internet for selfies, family photos, celebrity headshots, mugshots, yearbooks, and news photos to build massive image training datasets. If you've ever used Flickr, YouTube, Instagram or given a lecture at a conference these datasets might include you. If you're popular enough to be on Wikipedia there's an even higher chance your face has been used.
Although many of the datasets began as academic research projects, the MegaPixels website has found thousands of research papers linked to commercial organizations including SenseTime, Megvii/Face++, Hikvision, Amazon, Microsoft, IBM, Facebook, Google, Hitachi, Tencent; as well as dozens of research projects from defense and intelligence organizations including the National University of Defense Technology (CN) and IARPA (US). For a more complete analysis of each dataset, visit the MegaPixels research project.
The MegaPixels installation at Glassroom SF uses a custom face recognition system that does not share any data. Images are processed local and never transmitted.