Re:Extracting image metadata at scale
by TTPrograms on Monday March 21, 2016 @03:53PM
(#11331305)
They explain that not all of the images they want are faces in this case, so you'd have to train your own on "interesting regions" (though there is some work in that area). Part of the challenge in that case is generating all your labels for what the interesting regions are. This way they don't need to generate labels, at least.
[ Parent
| on HN ]
Youtube did something similar for 'interesting thumbnails' last year with deep nets (many uploaders do not specify a good thumbnail preview), and reported that it gave a nice performance boost.