Evaluation Data (Instance Segmentation)

COCO2017

#Metrics-segm mAP

Data description:

The COCO dataset is a large-scale object detection dataset funded by Microsoft in the United States. It encompasses 80 categories, with over 200,000 annotated images and 1,500,000 target instances. This dataset is sourced from a variety of complex daily scenarios, and its data distribution is relatively balanced. The training set, validation set, and test set contain 118,287, 5,000, and 40,670 images respectively. Since the annotations of the original test set are not publicly available, the original validation set is used as the test set.

Dataset structure:

Amount of source data:

The dataset is split into train(118,287), validation(5,000), test(5,000)

Amount of test data:

All 5,000 test examples from the source dataset test set.

Data detail:

Raw image：RGB image

instance annotation

Sample of source dataset:

Raw image：

annotation format：

[{ "image_id": int, "category_id": int, "segmentation": RLE, "score": float, }]

Citation information:

@article{2014Microsoft,
  title={Microsoft COCO: Common Objects in Context},
  author={ Lin, Tsung Yi  and  Maire, Michael  and  Belongie, Serge  and  Hays, James  and  Zitnick, C. Lawrence },
  journal={Springer International Publishing},
  year={2014},
}

Licensing information：

TERMS OF USE