What is COCO?


COCO is a new image recognition, segmentation, and captioning dataset. COCO has several features:

  • Object segmentation
  • Recognition in Context
  • Multiple objects per image
  • More than 300,000 images
  • More than 2 Million instances
  • 80 object categories
  • 5 captions per image
  • Keypoints on 100,000 people


Tsung-Yi Lin Cornell Tech
Genevieve Patterson Brown
Matteo Ruggero Ronchi Caltech
Yin Cui Cornell Tech
Michael Maire TTI Chicago
Serge Belongie Cornell Tech
Lubomir Bourdev UC Berkeley
Ross Girshick Facebook AI
James Hays Georgia Tech
Pietro Perona Caltech
Deva Ramanan CMU
Larry Zitnick Facebook AI
Piotr Dollár Facebook AI



NEWS: We are pleased to announce the COCO 2016 Detection and Keypoint Challenges!
Results will be announced at the ImageNet and COCO Visual Recognition workshop at ECCV. Note: Competition deadlines have been extended to 09/16/2016.