COCO

COCO is a new recognition, segmentation, and captioning dataset. COCO has several features: Object segmentation Recognition in Context Multiple objects per More than 300,000 images More than 2 Million instances 80 object categories 5 captions per Keypoints on 100,000 people

Dataset Attributes

TasksDetection

CategoriesHumans, Objects, Vehicles, Nature, Furniture, Animals, Food, Kitchenware, Electronics

SensorRGB Camera

Class Labels

persondogcowtraincarmotorbikechairsofabottle