search

Understanding-Visual-Datasets / VisDiff

Official implementation of "Describing Differences in Image Sets with Natural Language" (CVPR 2024 Oral)
https://understanding-visual-datasets.github.io/VisDiff-website/
Apache License 2.0
105 stars 12 forks source link

Extra config yaml request #8

Open LeoooooMM opened 4 months ago

LeoooooMM commented 4 months ago

Hello, thank you for your wonderful work! In order to reproduce the results in your paper, I wonder if you will provide the config of three data sets in the near future

lisadunlap commented 4 months ago

We're glad you like our work! If you look under the sweeps folder you should be able to reproduce the results for the 3 datasets :)

Let us know if you have any more questions!

LeoooooMM commented 3 months ago

Thank you for your prompt reply! I will close this issue

LeoooooMM commented 3 months ago

Excuse me, in order to get the results of the application in the paper, I guess it is necessary to first download Pictures in official website, including Lamen and Imagenet v1v2 (if correct, Imagenet v1 should be downloaded at https://www.image-net.org/download.php. And where v2 can I download the original picture?) As for clip_vs_resnet application, I would like to ask where is the imagenet-v2.csv file and whether the results of the paper can be obtained by just running main.py file in application/clip_vs_resnet?