Request for Evaluation Code of Tables 8/11/12/13

[Super-lyh-neversleep]

Hi authors,

Thank you for releasing the MoBGS code and for the great work.

I am trying to reproduce the supplementary evaluation results on the Stereo Blur dataset, especially the following tables:

• Table 8: Per-scene MUSIQ of full region
• Table 11: Per-scene LPIPS of dynamic region
• Table 12: Per-scene tOF of dynamic region
• Table 13: Per-scene PSNR of dynamic region

However, I could not find the exact evaluation scripts for these tables in the released code. The current metrics.py seems to mainly compute full-image PSNR / SSIM / LPIPS / tOF, and I could not find the exact dynamic-region evaluation code.

Would it be possible to release the scripts used to compute Tables 8, 11, 12, and 13?

In addition, I would like to ask one specific detail about the dynamic-region evaluation:

During dynamic-region evaluation, did you apply dilation to the provided dynamic masks?

For example, DyBluRF appears to use a dilation operation like:

skimage.morphology.dilation(motion_mask, skimage.morphology.disk(2))

Did MoBGS use the same disk(2) dilation or any other dilation operation when computing the dynamic-region LPIPS / tOF / PSNR in Tables 11/12/13?

If yes, could you please specify the dilation radius/kernel used?

Thank you very much!