Data for evaluation

[leezythu]

The work is great!
I'd like to know where I can get the subset of datasets for evaluation used in the paper?

[jwyang]

glad you like it! We will release the evaluation datasets shortly this week!

[KainingYing]

Thanks for your great work. Look forward to data for evaluation.

[jwyang]

@leezythu, @KainingYing

thanks again for your interest in our benchmark. Please find the first version of our SoM-Bench here:

https://github.com/microsoft/SoM/tree/main/benchmark

Please note that we still working on the compilation of the ground-truth annotations for COCO, ADE20K and the automatic evaluation code now that GPT-4V API has been released. For now, you can try flickr30k and refcocog first.

thanks!

[KainingYing]

Great work! Thanks :)