PROJET AUTOBLOG


shaarli-Links

Site original : shaarli-Links

⇐ retour index

COCO - Common Objects in Context

mercredi 7 septembre 2016 à 23:38
«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
»
(Permalink)