Every year we take a group photo at Akademy and then me or one of the Kennies manually marks up the faces so people can tag them and we can know who we all are and build community. This is quite old school effort so this year I followed a mangazine tutorial and made Akademy Group Photo Automator to do it. This uses an AI library called face_recognition to do the hard work and Docker to manage the hard work and spits out the necessary HTML. It was a quick attempt and I’m not sure it did much good in the end alas. The group photos tend to be quite disorganised and whoever takes it upon themselves to direct it each year makes basic mistakes like putting everyone on a flat stage or making everyone wave their hands about which means many of the faces are half covered and not recognised. And it seems like the library is not a fan of glasses. It also outputs rect coordinates rather than circle ones which ment Kenny had to do many adjustments. Still it’s an interesting quick dive into a new area for me and maybe next year I’ll get it smoother.
