The remarkable capabilities of the Segment Anything Model (SAM) for tackling image segmentation tasks in an intuitive and interactive manner has sparked interest in the design of effective visual prompts. Such interest has led to the creation of automated point prompt selection strategies, typically motivated from a feature extraction perspective. However, there is still very little understanding of how appropriate these automated visual prompting strategies are… See the full description on the dataset page:
https://huggingface.co/datasets/gOLIVES/SAM_PointPrompt_Dataset.
