PROJECT TITLE :

Human-Machine CRFs for Identifying Bottlenecks in Scene Understanding

ABSTRACT:

Recent trends in image understanding have pushed for scene understanding models that jointly reason regarding varied tasks such as object detection, scene recognition, shape analysis, contextual reasoning, and local appearance primarily based classifiers. In this work, we have a tendency to are curious about understanding the roles of these totally different tasks in improved scene understanding, in explicit semantic segmentation, object detection and scene recognition. Towards this goal, we have a tendency to “plug-in” human subjects for each of the various parts in a very conditional random field model. Comparisons among numerous hybrid human-machine CRFs offer us indications of how a lot of “head room” there's to improve scene understanding by focusing analysis efforts on varied individual tasks.