Skip to content

Adaptation of the provided code to the Base+ version for Shadow Detection(ISTD) #139

Description

@ZeeshanMasood786

Hi,
I was trying to adapt the provided code SAM 2.1 large model to the sam2.1 Base+ model for the shadow detection on ISTD dataset. I made changes to heiradet.py, image_encoder.py and sam.py to changes the hardcoded parameters of large model to Base+ model. After training several time with minor changes I can only achieve the BER to 588 to 5.95. I wanted to ask if someone else has tried to adapt the provided code for other models on Shadow Detection or other datasets, let me know. I wanted to understand whether this is becuase of the capacity of the Base+ model or the inefficient adaptation that i have done so far. Lets connect someone else has trained in the same way and if he/she got some good BER results.
Thanks & Regards.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions