search

MichalisLazarou / iLPC

accepted at ICCV 2021
MIT License
20 stars 2 forks source link

pkl_dataset.transform_cfg import transforms_test_options, transforms_list #1

Open allenhaozhu opened 2 years ago

allenhaozhu commented 2 years ago

sorry, I cannot run the demo you provide. In test_arguments.py, from pkl_dataset.transform_cfg import transforms_test_options, transforms_list seems not exist in your source codes. Thanks for your attention.

MichalisLazarou commented 2 years ago

Hey Allen, apologies for that I have added the code can you please let me know if it works now?

allenhaozhu commented 2 years ago

works well! Thank you very much. I hope I can catch CVPR and cite your paper.

MichalisLazarou commented 2 years ago

Hope it helps your work. All the best for your paper, good luck with CVPR!

Doflamingo-swj commented 1 year ago

sorry, I cannot run the demo you provide. In test_arguments.py, from pkl_dataset.transform_cfg import transforms_test_options, transforms_list seems not exist in your source codes. Thanks for your attention.

Hello, maybe the question I asked is irrelevant to your question. How did you get the author's following ckpt file? Did you apply on Google? I tried for a long time. Thank you for your attention. "WRN-28-10: download and extract [checkpoints] folder from here: checkpoints"

Doflamingo-swj commented 1 year ago

Hey Allen, apologies for that I have added the code can you please let me know if it works now?

Hello, this may be impolite, but I didn't see a reply in the issue proposed by another person, so here I ask: I submitted an application for accessing ckpt on Google drive, can you pass it? I like your work very much, thank you ！

allenhaozhu commented 1 year ago

Hey Allen, apologies for that I have added the code can you please let me know if it works now?

Hello, this may be impolite, but I didn't see a reply in the issue proposed by another person, so here I ask: I submitted an application for accessing ckpt on Google drive, can you pass it? I like your work very much, thank you ！

Technically, I sucessfully make it runable. But I cannot obtain the reasonable performance. So I only use the code for time comparison (very slow) in our paper. You could follow our work in CVPR2022 (EASE: Unsupervised Discriminant Subspace Learning for Transductive Few-Shot Learning). The code is very simple and fast.

MichalisLazarou commented 1 year ago

Sounds good, what performance do you get when you say you dont get the same performance? Especially wrn should be very similar if you use the features extracted +-0.5