-
Notifications
You must be signed in to change notification settings - Fork 20
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
About 17w training samples in papers #65
Comments
hello, somewhat we have two version of data, a initial version of 220000 samples while the 170000 shapes has a clip score above 28. You could use text-rendering clip score to filter the 220000 shapes but overall the 220000 shapes are all for training. |
Thank you for your reply~ |
|
Thank you very much~ |
Hello, Can you provide me with the download link again? The current link has expired. Can you provide me with the division of training, testing and validation sets? |
https://virtualbuy-devo.oss-cn-hangzhou.aliyuncs.com/muyuan/CVPapers/tmp/valid_paths_v4_cap_filter_thres_28_catfilter19w.json?OSSAccessKeyId=LTAI5tAGAZPVv9b26UUBYDuk&Expires=1746752381&Signature=KCk0w35rXKmgaB1cO9Tr8nb1H0I%3D |
The paper mentioned the use of the gobjaverse dataset, which has about 170,000 shapes. However, the current gobjaverse index json contains 260,000 shapes. After removing more than 40,000 poor quality shapes, there are still 220,000 shapes. How were the extra 50,000 shapes screened out?
The text was updated successfully, but these errors were encountered: