-
Notifications
You must be signed in to change notification settings - Fork 25
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
Questions about refcocog datasets split. #7
Comments
Thank you for pointing out this typo error. The descriptions in the paper might be 'umd' instead of 'unc'. In fact, there are 104,560 expressions in the original RefCOCOg (google), which only contains 'train' and 'val' split. But when it was re-splitted in the 'umd' partition, some expressions are filtered out. |
Thanks for quick reply. However, when I count the right number with umd split.
Anyway, Thanks very much to solve my concern. |
Hello, thanks for your wonderful work.
![image](https://user-images.githubusercontent.com/27133144/93293639-06cb1780-f81b-11ea-9f9f-95fb372bf3c4.png)
![image](https://user-images.githubusercontent.com/27133144/93293733-3bd76a00-f81b-11ea-823a-0622e258b470.png)
![image](https://user-images.githubusercontent.com/27133144/93293795-6de8cc00-f81b-11ea-86e5-871b9cfe4772.png)
In your paper, you claim that you use the UNC partition for refcocog dataset.
However, in your DATA_PREP only has the umd split
And then if I use the default umd split.
The total expression for train + val + test = 95010, which is much fewer than the original expressions (104560)
Can you help to figure out what's wrong with it? Thanks very much!
The text was updated successfully, but these errors were encountered: