PhD student develops algorithm to predict how popular a photo will be online


Even before selfies became popular, I was taking self-portraits with either mirror shots or photos of my feet or shoes wherever I am. That was before when I was younger and adventurous. But right now, I am too lazy to bring my camera gears and just depend on my trust old iPhone 4. Unlike before, I would bring a handful of cameras: my camera phone, my DSLR, a digicam for shooting videos, and my Flip camera for HD videos. Crazy, right?

Today, I easily get annoyed seeing my Facebook or Instagram feed with super vain selfie shots. I don’t understand this, really. Do these people find themselves so attractive that sharing their photos make them happy? I know time will come that people will get tired of selfies. But for the really vain ones, there is a way for science to help you with your obsession and vanity. (Is there an end to this?!)

I’m being sarcastic here but Aditya Khosla, an MIT PhD student, managed to study data using about 2.3 million Flickr photos. Khosla developed an algorithm that predicts popularity of a photo shared online. Accuracy depends on image content, social influence, colour, tags, and types of objects included in the photo. One can determine how people will look, like, or share a photo.

The study identified that sexy shots are sure winners, included are bikinis, bras, and mini-skirts being displayed prominently. It’s not really surprising because more people, especially the men, are drawn to photos that are sexy.

So what’s the value of this study? That people need to do post more sexy selfies? I don’t think so. Khosla is aiming to modify pictures with edits automatically to make them popular on the web. It’s possible. But yes, this one will benefit those who like taking selfies.

(Seriously, enough!!!)

[via Techcrunch]

Related Posts