Sama-COCO

Sama-COCO数据集是由Sama Montréal, Québec机构重新标注的MS-COCO数据集，旨在通过精确的标注提高数据质量。该数据集包含123,287张图像，重点在于精确的像素级标注和分解人群实例。创建过程中，多达500名标注者参与，遵循严格的标注指南，确保标注的精确性。Sama-COCO数据集在自动驾驶和图像预训练等领域有广泛应用，特别适用于需要高精度实例分割的任务。

The Sama-COCO dataset is a re-annotated version of the MS-COCO dataset developed by Sama Montréal, Québec, aiming to enhance data quality through precise annotations. It contains 123,287 images, with a focus on accurate pixel-level annotations and disaggregated crowd instances. Up to 500 annotators participated in its creation, following strict annotation guidelines to ensure annotation precision. The Sama-COCO dataset has widespread applications in fields such as autonomous driving and image pre-training, and is particularly suitable for tasks requiring high-precision instance segmentation.