no code implementations • NeurIPS 2009 • Tomasz Malisiewicz, Alyosha Efros
The use of context is critical for scene understanding in computer vision, where the recognition of an object is driven by both local appearance and the objects relationship to other elements of the scene (context).
no code implementations • NeurIPS 2009 • Bryan Russell, Alyosha Efros, Josef Sivic, Bill Freeman, Andrew Zisserman
In contrast to recent work in semantic alignment of scenes, we allow an input image to be explained by partial matches of similar scenes.