test set for region caption on VG dataset

[liweiyangv]

Thank you for your work.
I am wondering what the split of the test set you used for the region-level caption task on the VG dataset. Is the test split used by paper Glamm?

[SunzeY]

Yes, this is also aligned with GPT4ROI

[liweiyangv]

Thanks for your reply. I am writing to confirm that you use a total 5000 bbox for testing.