About the vanilla SAM?

[qianyuli123]

Thank you very much for open-sourcing your work. I would like to ask that in the experimental part of your work, there is a comparison with the initial SAM. It is mentioned that manual point annotations were provided for each gland in the CRAG dataset and the segmentation results occupying half of the image area were filtered. May I ask if the point annotations or the relevant code in this process have been saved? Could you please share them? Thank you very much.

[jingweizhang-xyz]

The point annotations are generated using the ground truth. This should be the code we used to generate dot annotations, not 100% sure. The vanilla SAM code should contain some scripts to generate segmentation masks given dot annotations, I remember.
vanilla_sam_gen_dot.zip