Common Objects in Context

Common Objects in Context (COCO) is a dataset of photos used for object detection. It includes a "person" class with 59,144 unique photos from Flickr. The dataset was published in 2014 and contains 328,000 total images. The dataset is likely the most widely used benchmarking dataset in the world for object detection.

It is even referenced in slides presentation by the Office of Directory of National Intelligence, noting the objects that are relevant to the IC (Intelligence Community), including Person, Car, Truck, Backpack, and Knife.

The full extent of this dataset has not yet been fully analyzed due to the massive usage. Analysis is still in progress.

COCO Copyright Distribution

COCO Creative Commons License Distribution

COCO Image Upload Year Distribution

Top 10 COCO Image #Tags

Top 10 Geocoded Cities COCO

Citing This Work

If you reference or use any data from the Exposing.ai project, cite our original research as follows:

@online{Exposing.ai,
  author = {Harvey, Adam. LaPlace, Jules.},
  title = {Exposing.ai},
  year = 2021,
  url = {https://exposing.ai},
  urldate = {2021-01-01}
}

If you reference or use any data from COCO cite the author's work:

@article{Lin2014MicrosoftCC,
    author = "Lin, Tsung-Yi and Maire, M. and Belongie, Serge J. and Hays, J. and Perona, P. and Ramanan, D. and Doll{\'a}r, P. and Zitnick, C. L.",
    title = "Microsoft COCO: Common Objects in Context",
    journal = "ArXiv",
    year = "2014",
    volume = "abs/1405.0312"
}

References