-
One-shot Scene Graph Generation
One-shot scene graph generation task, utilizing rich prior knowledge from Visual Genome and ConceptNet. -
COCO-stuff and Visual Genome
The dataset used in the paper is COCO-stuff and Visual Genome. COCO-stuff is a dataset of 164K images with pixel-level stuff annotations, and Visual Genome is a dataset of... -
Visual Genome
The Visual Genome dataset is a large-scale visual question answering dataset, containing 1.5 million images, each with 15-30 annotated entities, attributes, and relationships.