Request for the code of evaluation metrics (e.g., relative confidence drop, localization score)

Hi authors,
I’m deeply interested in your work Finer-CAM: Spotting the Difference Reveals Finer Details for Visual Explanation published in CVPR 2025. The method’s design for fine-grained visual explanation is really insightful!
I noticed that the paper provides detailed visualization results and quantitative metrics (like relative confidence drop (RD), deletion AUC, and energy-based pointing game for localization), but the released code repo doesn’t include the implementation of these evaluation metrics yet.
I’m currently working on reproducing and comparing fine-grained XAI methods, and the standard, official implementation of these metrics would be crucial for ensuring the fairness and accuracy of my experiments. Could you please consider open-sourcing the code for these evaluation metrics?
Thank you so much for your contributions to the community and your kind assistance!
Best regards。

The issue names no target files, tests, or entry points. Start by comparing the paper's definitions of relative confidence drop, deletion AUC, and energy-based pointing game with the repository's existing evaluation flow; done means reproducible implementations of the requested metrics are available.

Written by the indexing model from the issue text.