Can the vis dataset extract only the mask of a specific class such as person?

[ShellyLingling]

Is the format of annotation exactly the same as the coco dataset? @linjieyangsc Can you give me some references?