zerovl / ZeroVL

[ECCV2022] Contrastive Vision-Language Pre-training with Limited Resources
MIT License
43 stars 5 forks source link

train_anno.csv and val_anno.csv #2

Open haoshuai714 opened 2 years ago

haoshuai714 commented 2 years ago

could you provide the train_anno.csv and val_anno.csv file?

zerovl commented 2 years ago

Yes, could you leave your email address or something else, and I will send files to you. BTW, i need to clarify that you would better create the annotation files by yourself, since the images are downloaded from urls and some urls could be 404. Therefore, downloaded images and filenames could be different for different people. To solve this problem, the texts of images could be used as the main key to filter training samples.

zerovl commented 2 years ago

https://pan.baidu.com/s/1EIPztHFLjeV_03AbnlvDZQ?pwd=43gn you will find annotation files of coco, f30k and sbu in this url. I will update annotation files of other datasets in the following days.

haoshuai714 commented 2 years ago

Thanks! what is the image_id and caption_id in the csv files? How to get it?

haoshuai714 commented 2 years ago

The sub dataset csv file only contain image and caption. In the pretraining phase, the csv file only have image and caption ok?

zerovl commented 2 years ago

Hi. The image_id and caption_id can be randomly given. I rename the image files, so the image_id is the same to the image file name. As I said before, you would better create annotation files by yourself, since the image filenames can be different for different people. It could be flexible to create the annotation files by yourself for training with your own datasets.

For pretraining, only providing image and caption is okay, since image_id and caption_id are used for evaluation. For evaluation, image_id and caption_id are both necessary. From: @.> Date: 2月28日 (周一) 22:00 Subject: [External] Re: [zerovl/ZeroVL] train_anno.csv and val_anno.csv (Issue #2) To: @.> Cc: @.>, "Comment"< @.>

The sub dataset csv file only contain image and caption. In the pretraining phase, the csv file only have image and caption ok?

— Reply to this email directly, view it on GitHub https://github.com/zerovl/ZeroVL/issues/2#issuecomment-1054319706, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXLIQK7KB3LQKZCKNMY7YXLU5OBZVANCNFSM5PMGDLWA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub . You are receiving this because you commented.Message ID: @.***>

haoshuai714 commented 2 years ago

Thanks!

zerovl commented 2 years ago

my pleasure! From: @.> Date: 2月28日 (周一) 23:00 Subject: [External] Re: [zerovl/ZeroVL] train_anno.csv and val_anno.csv (Issue #2) To: @.> Cc: @.>, "Comment"< @.>

Thanks!

— Reply to this email directly, view it on GitHub https://github.com/zerovl/ZeroVL/issues/2#issuecomment-1054363037, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXLIQKZN2K33XIYL2PGMWGDU5OHA7ANCNFSM5PMGDLWA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub . You are receiving this because you commented.Message ID: @.***>

haoshuai714 commented 2 years ago

Hello! when I try to run the code. I have a root problem:ModuleNotFoundError: No module named 'zerovl'No module named 'zerovl': No module named 'zerovl'? if you have this problem?

zerovl commented 2 years ago

I think it could be caused by the absolute import. Could you try to include the repo path to your PYTHONPATH variable? For instance: export PYTHONPATH='path_to_the_code'.

haoshuai714 commented 2 years ago

yes, it ok. But, have a new bug:PermissionError: [Errno 13] Permission denied: '/home/tiger'; what is cache_dir=f'/home/tiger/.cache/torch/hub/checkpoints/{cfg.model.text_encoder.tag}')?

zerovl commented 2 years ago

Sorry, it is a mistake. It is a cache folder on my machine to save temporary files. Could you please replace the path to any path on your machine, and try again?

zerovl commented 2 years ago

For instance, you could replace '/home/tiger' to '~' directly.

haoshuai714 commented 2 years ago

OK, I try to do it. if you have WeChat? my WeChat id: haoshuai714

zerovl commented 2 years ago

yeah, i will add you.