They are feeding 10M frames from random YouTube videos, 1 frame per video. Only ... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

yaroslavvb on June 22, 2012 | parent | context | favorite | on: Building high-level features using large scale uns...

They are feeding 10M frames from random YouTube videos, 1 frame per video. Only 3% of 60x60 patches from those frames contained faces

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact