BBC News reported recently about a photo tool in development that would turn anyone’s bad photos into brag-worthy ones. The algorithm, created by James Hays and Alexei Efros from Carnegie Mellon University, scans a database of 2.3 million images taken from Flickr.
Before: The roof blocks the view
After: The roof is removed and boats are added from similar shots found online
The program would select and remove the unwanted item, search the database and pick out the closest 200 photos. From there it checks for elements that are the right size and color to fill the hole. Useful parts of the best 20 images are cropped and added in.
The team is also working on a project to create a library of clip art from photos on image sharing sites.
These programs could be useful, or people could just take better photos to begin with.