3 datasets found

The V-COCO dataset is a large-scale visual semantic graph dataset, containing 4,946 images annotated with object and predicate labels to form a scene graph.

The GQA dataset is a large-scale visual semantic graph dataset, containing 8,208 images annotated with object and predicate labels to form a scene graph.

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.