search

harrywuhust2022 / Adv-Crowd-analysis

ACM Multimedia 2021, Poster
MIT License
9 stars 1 forks source link

Is this the latest code? #1

Open yijie422 opened 3 years ago

yijie422 commented 3 years ago

Hi, i'm interest in your work, but i can't run your code. Is this the latest? I meet some problems, such as: file "csr_attack.py", line 298, why use CrossEntorpyLoss? And line 308, there is an obvious SyntaxError, how to solve it?

harrywuhust2022 commented 3 years ago

Thanks for your issue! the code is not the latest. The paper has recently finished the rebuttal phase of ACM MM2021 (but still under review), we will complete the repo after the notification. This is the draft code.

yijie422 commented 3 years ago

ok, thank you, looking forward to your updates.

yijie422 commented 2 years ago

Thanks for your issue! the code is not the latest. The paper has recently finished the rebuttal phase of ACM MM2021 (but still under review), we will complete the repo after the notification. This is the draft code.

Congratulations, will the code be updated recently?

harrywuhust2022 commented 2 years ago

Thanks for your issue! the code is not the latest. The paper has recently finished the rebuttal phase of ACM MM2021 (but still under review), we will complete the repo after the notification. This is the draft code.

Congratulations, will the code be updated recently?

Thank you! I have tried my best to find the code file on the computer, and the current version is the latest version I can find (I have uploaded). Because this project involves cooperation with several collaborators, some changes cannot be presented (we thank for your understandings!). I am surprised that you are actually interested in the robustness of crowd counting. From my perspectives, the most important problem in the robustness of regression modeling (not just the crowd counting models) is how to define the robustness evaluation metrics? I have already given some answers but I think this is still the open problem (actually, researchers still use the specific evaluation metrics on specific tasks like miou in semantic segmentations). There is already an article discussing this field (https://arxiv.org/abs/1911.11484) maybe you could discuss the robustness problem with him, and my current research interest has turned to sparse learning (lottery ticket hypothesis). After this answer, I am busy with the summer internship work and will not reply to the questions raised in this project, thanks for the understandings and good luck!