search

Alibaba-MIIL / ASL

Official Pytorch Implementation of: "Asymmetric Loss For Multi-Label Classification"(ICCV, 2021) paper
MIT License
732 stars 102 forks source link

annotations of NUS WIDE #20

Closed JiaweiZhao-git closed 3 years ago

JiaweiZhao-git commented 3 years ago

Hi, I download the NUS WIDE from the url presented in paper's appendice https://drive.google.com/file/d/0B7IzDz-4yH_HMFdiSE44R1lselE/view..

However, it seems there are only images without annotations and your train/test split list, could you please upload annotations and split list?

Thanks!

mrT23 commented 3 years ago

to the best of my recollection, we obtained the labels from the official nus site. some images had no labels, so they were removed from the 70-30 split anyway, I updated the model_zoo with the link https://github.com/Alibaba-MIIL/ASL/blob/main/MODEL_ZOO.md

notice that this is a major contribution of our article to the community, that should be properly recognized and cited. obtaining and properly labeling nus-wide was hard. no previous article has made even a remote attempt to share a well-labeled and organized version of nus-wide.