kkhoot / PAA

A PyTorch implementation of the paper `Probabilistic Anchor Assignment with IoU Prediction for Object Detection` ECCV 2020 (https://arxiv.org/abs/2007.08103)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Pretrained Models.

Joker316701882 opened this issue · comments

Hi. @kkhoot

Sorry for bothering you. We would like to cite your work in our work and run some extra tests on PAA. Is it possible for you to upload the following pre-trained models?

1). PAA_X_101_64x4d_FPN_2x
2). PAA_dcnv2_X_101_64x4d_FPN_2x

Bests!

Hi @Joker316701882 , I don't have those models, so I need to train them again. When is your due date for submission?

@kkhoot
No worry. I already start to train these two models with your code! Thank you for your kind reply : )

Great that you are able to train them yourself! Sorry for this inconvenience though.

@Joker316701882 JFYI I trained PAA_X_101_64x4d_FPN_2x and added the model link on READEM.md. This models shows 46.4 mAP on COCO val and 46.8 mAP on test-dev. Hope this helps.

@kkhoot
Thank you very much!

@Joker316701882 PAA_dcnv2_X_101_64x4d_FPN_2x model link also has been updated. It shows 48.8 (val) and 49.2 (test-dev).