COCO (Common Objects in Context), first published as Microsoft COCO, is a large image dataset used to train and test models that find, outline and describe objects in photos.
In plain words
COCO collected photos of busy everyday scenes, where objects sit in their normal surroundings. Crowd workers labelled each object separately, tracing its exact outline, not just drawing a box. That makes COCO useful for object detection, segmentation and image captioning.
A quick example
Imagine a photo of a kitchen. In COCO, each cup, chair and person in it is labelled separately, with a shape tracing its edges. A detection model is scored on how well its predicted boxes and outlines match those labels.
Where you'll meet it
- Benchmarking object detection models.
- Training and testing image segmentation and captioning.
Related
Sources
- PaperMicrosoft COCO: Common Objects in Context, Lin et al. (arXiv) · read 29 Sept 2026
- Docscoco, TensorFlow Datasets catalog · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.