MSCOCO

E899057

benchmark dataset computer vision dataset image captioning dataset image dataset

MSCOCO is a large-scale benchmark dataset of everyday images with rich object annotations and human-written captions, widely used for training and evaluating computer vision and image captioning models.

Try in SPARQL Jump to: Statements Referenced by

Statements (51)

Predicate	Object
instanceOf	benchmark dataset ⓘ computer vision dataset ⓘ image captioning dataset ⓘ image dataset ⓘ
abbreviation	COCO NERFINISHED ⓘ
developedBy	Microsoft ⓘ Microsoft Research NERFINISHED ⓘ
domain	computer vision ⓘ
evaluationMetric	BLEU NERFINISHED ⓘ CIDEr NERFINISHED ⓘ METEOR NERFINISHED ⓘ ROUGE-L NERFINISHED ⓘ SPICE NERFINISHED ⓘ mAP@[.5:.95] ⓘ mean Average Precision ⓘ
fullName	Microsoft Common Objects in Context NERFINISHED ⓘ
hasAnnotationType	human-written captions ⓘ image-level labels ⓘ keypoints ⓘ object bounding boxes ⓘ object categories ⓘ object segmentation masks ⓘ
hasModality	natural images ⓘ
hasProperty	context-rich images ⓘ everyday scenes ⓘ large-scale ⓘ multiple objects per image ⓘ rich annotations ⓘ
hasSplit	test set ⓘ training set ⓘ validation set ⓘ
hasTask	COCO Captioning Challenge NERFINISHED ⓘ COCO Detection Challenge NERFINISHED ⓘ COCO Keypoints Challenge NERFINISHED ⓘ COCO Segmentation Challenge NERFINISHED ⓘ
isWidelyUsedIn	academic research ⓘ benchmarking vision models ⓘ
license	non-commercial research license ⓘ
releasedBy	Microsoft COCO Consortium NERFINISHED ⓘ
usedFor	deep learning research ⓘ human pose estimation ⓘ image captioning ⓘ instance segmentation ⓘ keypoint detection ⓘ model evaluation ⓘ object detection ⓘ object recognition ⓘ semantic segmentation ⓘ supervised learning ⓘ visual question answering ⓘ
website	https://cocodataset.org ⓘ

Referenced by (1)

Full triples — surface form annotated when it differs from this entity's canonical label.

Show and Tell: A Neural Image Caption Generator → usesDataset → MSCOCO ⓘ