HoNeYBiRD
Elite Member
- May 1, 2009
- 10,187
- 12,580
What makes an image popular? Can we predict the number of views an image will receive even before it is uploaded?
It appears MIT has the answer. Using a dataset of about 2.3 million images from Flickr, they say that they can reliably predict the normalized view count of images.
Popularity Demo:
Popularity API:
IMG_URL needs to be replaced with the direct link to your image.
Higher number means higher popularity. It's a bit slow unfortunately.
I've just read about this on a local news site i frequent. The author of the article claims that he's using it for weeks now and the results given by the tool seem to be accurate.
Image content is just one "ingredient" in making an image popular, social cues play a big role too. The above tool is only good for image content analysis obviously. If anyone's interested, these MIT guys have a not too long PDF study around the subject on their site where they combine image content and social cues to predict image popularity.
It appears MIT has the answer. Using a dataset of about 2.3 million images from Flickr, they say that they can reliably predict the normalized view count of images.
Popularity Demo:
Code:
http://popularity.csail.mit.edu/
HTML:
http://popularity.csail.mit.edu/cgi-bin/image.py?url=IMG_URLHigher number means higher popularity. It's a bit slow unfortunately.
I've just read about this on a local news site i frequent. The author of the article claims that he's using it for weeks now and the results given by the tool seem to be accurate.
Image content is just one "ingredient" in making an image popular, social cues play a big role too. The above tool is only good for image content analysis obviously. If anyone's interested, these MIT guys have a not too long PDF study around the subject on their site where they combine image content and social cues to predict image popularity.
