Scene completion.This is a toy implementation of Scene Completion Using Millions of Photographs for 6.870 Object Recognition and Scene Understanding.-This implementation uses a　of 54,000 256x256 images　the Oliva lab for which we already have gist-based nearest neighbors data.