Thank you very much for your outstanding work. In the paper, you demonstrated the fine-tuning performance on the refcoco dataset. May I ask if this was achieved by adding the refcoco dataset on top of the object365 dataset? If so, can I directly add the refcoco dataset to the stage2_with_object365.sh script? In this case, what is the repeat coefficient for refcoco? I greatly look forward to your reply!
Thank you very much for your outstanding work. In the paper, you demonstrated the fine-tuning performance on the refcoco dataset. May I ask if this was achieved by adding the refcoco dataset on top of the object365 dataset? If so, can I directly add the refcoco dataset to the stage2_with_object365.sh script? In this case, what is the repeat coefficient for refcoco? I greatly look forward to your reply!