Questions about the visual grounding task

[ander008]

The format of dataset about the visual grounding task is not described clearly in your repo. Can you give some clear instructions? Or, How should I fine-tune the model on my own dataset about visual grounding task.