Yangsenqiao / vida

[ICLR 2024] ViDA: Homeostatic Visual Domain Adapter for Continual Test Time Adaptation
MIT License
48 stars 4 forks source link

About segmentation codes #1

Open daeunni opened 1 year ago

daeunni commented 1 year ago

Hi, thanks for your impressive work! :) I really looked forward to releasing this code, so thank u for sharing!

btw, may I ask whether u can share segmentation codes also?

cht619 commented 10 months ago

Also looking forward to the codes of segmentation

Yangsenqiao commented 10 months ago

Thank you for your interest in our project. We intend to release our code once the paper has been accepted. You could download our segmentation codebase from here.

JimPlayboy commented 6 months ago

We are very interested in your article. However, we found that the link you provided is the Cotta method (Continual Test-Time Domain Adaptation), not the code corresponding to your article (VIDA: HOMEOSTATIC VISUAL DOMAIN ADAPTER FOR CONTINUAL TEST TIME ADAPTATION). When will the segmentation code corresponding your article be releasesd?

daeunni commented 6 months ago

@JimPlayboy Same here.

Yangsenqiao commented 6 months ago

Thank you for your interest in our work. We are in the process of organizing the code and plan to release it shortly after the deadline for ECCV :pray: :pray: . If you urgently need our segmentation code in the next few days, please send an email to yangsenqiao.ai@gmail.com, and I will send you the core code privately (however, some variable names may not be standardized, and may be somewhat messy.).

liujiaming1996 commented 4 months ago

Official Notice: Thank you for your interest in our work. We open-sourced the code for CiFAR100-CiFAR100C yesterday (2024.04.15). For segmentation code, you can quickly integrate it by following CoTTA's segmentation code framework and ViDA method code. For example, you can utilize entire CoTTA's framework, which includes the dataloader, model creation, training code, and evaluation part, and inject our proposed method into the SegFormer model.