a question about evalCOCO.m

[xiaojieli0903]

@ZheC hi: Thank you for providing the training and testing code. Recently when I use your testing code in matlab to test my trained model, I found that when you arrange the score of the subset, the code is: json_for_coco_eval(count).score = pred(j).annorect(d).annopoints.score*length(pred(j).annorect(d).annopoints.point); . I'm a little confused because I think it should be like this: json_for_coco_eval(count).score = pred(j).annorect(d).annopoints.score/length(pred(j).annorect(d).annopoints.point); which means the final score of a person should be the average of the scores of the existed visible parts? Could you please tell me the reason of using multiplication to get the final score?

[woodyhu]

Please do not send email to me with this thread

---

From: Xiaojie Li [email protected] Sent: Wednesday, December 20, 2017 8:07 PM To: ZheC/Realtime_Multi-Person_Pose_Estimation Cc: Subscribed Subject: [ZheC/Realtime_Multi-Person_Pose_Estimation] a question about evalCOCO.m (#129)

@ZheChttps://nam01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fzhec&data=02%7C01%7C%7C542609386adc4fcd354e08d547a24505%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636493684622291194&sdata=zEBwmdfu13U8JzlDOYeVWAb%2BCN7kPBO7MNkotSu7pec%3D&reserved=0 hi: Thank you for providing the training and testing code. Recently when I use your testing code in matlab to test my trained model, I found that when you arrange the score of the subset, the code is: json_for_coco_eval(count).score = pred(j).annorect(d).annopoints.score *length(pred(j).annorect(d).annopoints.point); . I'm a little confused because I think it should be like this: json_for_coco_eval(count).score = pred(j).annorect(d).annopoints.score /length(pred(j).annorect(d).annopoints.point); which means the final score of a person should be the average of the scores of the existed visible parts? Could you please tell me the reason of using multiplication to get the final score?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://nam01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FZheC%2FRealtime_Multi-Person_Pose_Estimation%2Fissues%2F129&data=02%7C01%7C%7C542609386adc4fcd354e08d547a24505%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636493684622291194&sdata=gSggVxItH6HSVB9t9ts0p7yUbYc2dzsgp%2F0GlTWMxtA%3D&reserved=0, or mute the threadhttps://nam01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAQK35cTiphPifI2pI8jKeJlSEKmbZxpmks5tCPiMgaJpZM4RITOP&data=02%7C01%7C%7C542609386adc4fcd354e08d547a24505%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636493684622291194&sdata=vLBt4X8kAiuycfDTiSd5a3ONXewtkxT3kZnfX9Rha24%3D&reserved=0.