Regarding results.pkl.json file for SIPMASK-VIS

[gauravmunjal13]

Hi Team, I hope you're doing well! I am reaching out to you regarding the interpretation of the results file, results.pkl.sjon for SIPMASK-VIS. There are multiple detections for each category on a single video. Presently, I am taking the one with the highest score. Is that correct? Here are the values:

video_id:1, category_id:1, score:0.2766743004322052
video_id:1, category_id:2, score:0.17311915755271912
video_id:1, category_id:1, score:0.04480154812335968
video_id:1, category_id:2, score:0.050576336681842804
video_id:2, category_id:1, score:0.21141134202480316
video_id:2, category_id:2, score:0.10147302597761154
video_id:2, category_id:2, score:0.06389215588569641
video_id:3, category_id:1, score:0.26379796862602234
video_id:3, category_id:2, score:0.09652866423130035
video_id:3, category_id:2, score:0.058702562004327774
video_id:3, category_id:2, score:0.06718862056732178
video_id:3, category_id:1, score:0.03977232053875923
video_id:4, category_id:1, score:0.12127449363470078
video_id:4, category_id:2, score:0.19678737223148346
video_id:4, category_id:1, score:0.18680523335933685
video_id:4, category_id:2, score:0.04028315097093582

Thank you!

[JialeCao001]

@gauravmunjal13 It seems mean value. Please refer to https://github.com/JialeCao001/SipMask/blob/d1b68d545383114fe43fe247e65eddc60125cfb5/SipMask-VIS/mmdet/core/evaluation/coco_utils.py#L205