Common Objects in Context (COCO) is one of the most common large-scale image datasets for the evaluation of state-of-the-art computer vision models. COCO dataset contains image annotations in 80 categories, with over 1.5 million object instances.
COCO’s 5 annotation types aid in the following operations:
- Object detection
- Keypoint detection
- Semantic segmentation
- Panapotic segmentation
- Image captioning
You can create a separate JSON file for training, testing, and validation purposes.
Annotation Type: instance segmentation, object detection
Created By: COCO Consortium
Publish Date: September 01, 2017
License: CC BY 4.0
Dataset Size: 287,135 images