Question about Table 3 in the paper — What type of mask is used for 40–50% occlusion on Places2? #151
Description
Hi, thank you for the great work!
I have a question regarding Table 3 in your paper, which presents quantitative comparisons for context-aware image inpainting on the Places2 dataset under 40–50% mask ratios.
Could you please clarify:
What type of masks were used for the 40–50% occlusion setting?
Were they random rectangles, semantic masks, free-form, or something else?
Understanding this would help us align our evaluation with yours more accurately.
Thanks in advance for your time!
Best regards,