Scene Parsing Using Scene Attributes As Global Features

Data-driven methods have been proven very effective for the task of scene parsing. A crucial step in these methods is to retrieve a set of visually similar scenes from existing image collections for the query image according to certain global scene representations. In this work, we incorporate scene attributes into data-driven scene parsing systems as global scene features. We show that when used as global features, our compact attribute-based scene representation can compete with or improve on traditional low-level scene representations for the task of scene parsing and scene retrieval in general.