Format of result.json

[zlmlaker]

Thanks for the great work. Just a little confusion here. What is the format of the result.json? Is it supposed to be the same format of the name_of_input_pose.json? I know name_of_input_pose.json should be the detection result from another model when running inference. However, is it right that when training for COCO, the result.json should be part of the annotation file person_keypoints_train2017.json except for the format difference?

[mks0601]

Please see here, which tells what result.json is.
You can download result.json at here