Skip to content
This repository has been archived by the owner on Nov 21, 2023. It is now read-only.

What does mask r-cnn's AP, AP50, AP70 mean? #55

Closed
bughunter99 opened this issue Jan 27, 2018 · 2 comments
Closed

What does mask r-cnn's AP, AP50, AP70 mean? #55

bughunter99 opened this issue Jan 27, 2018 · 2 comments

Comments

@bughunter99
Copy link

Hi.

I'm novice on r-cnn.There are term AP, AP50, AP75 on mask r-cnn paper. 50, 75 is small postfix, but I can't make it small, sorry. Anyway the paper says it is averaged over IOU thresholds. For AP50, only candidates over 50% region comparing ground truth are counted, and for AP75 only candidate over 75% are counted. Then what is empty AP? I knew AP use 70% IoU threshold, but it wouldn't. Because just AP is lower then AP75.

And additionally, there are another terms I don't understand well. They are APs, m, L. I know they mean small, medium, large. Then is there any criteria how big they are. Just saying small, medium, large makes me little bit confused.

Thanks in advance!

@ppwwyyxx
Copy link
Contributor

See coco home page: http://cocodataset.org/#detections-eval

@xQsM3
Copy link

xQsM3 commented Jan 20, 2021

which confidence score (or which range) is used for AP computation? I cannot find it at the coco home page you posted here. thanks for help

EDIT:
the guys from COCO confirmed that there is no confidence score range, instead, the 100 top scored detections are considered, no matter how low their confidence is

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants