Despite beingmore unsafe than swimming in shark - overrun waters , people ( and animals ) are still very much engage in selfie culture . Capturing the perfect ego - portrait can be more difficult than apprehend Moby - Dick , but thanks to Stanford computer science grad bookman Andrej Karpathy , there may be promise .
According toFast Company , Karpathy educate something call up a " convolutional neural electronic connection " ( which he explains in - depthon his blog ) to separate between honest and defective selfies . He first collected jillion of photograph that used the # selfie hashtag and narrowed them down to a more manageable two million selfie sampling , making trusted that each photo had at least one face . The photo were then ranked based on how many like they get , and that exfoliation was " ascertain " for the routine of followers each photographer had .
" Best 100 out of 50,000 selfies , as judged by the Convolutional Neural internet . " - Andrej Karpathy
After feed the datum into the ConvNets , Karpathy made observationsbased on thetop 100 selfies . To take the perfect selfie you should be a woman with long hair , your oversaturated face should engage at least one - third of the frame with your os frontale cut off , and there should be a filter and border add . " A just share of the unevenness between what makes a in force or bad selfies can be explained by the panache of the look-alike , " Karpathy said of his findings , " as opposed to the raw attractiveness of the person . "
It is interesting to note ( which Karpathy does not include in his observance ) that none of the women in the top 100 have dour skin , only two out of 100 fatigue eyeglasses , and none of them are dressed in bright coloring material .
Head over toKarpathy ’s blogto pick up more about the convolutional neural web and the results of his experimentation .
