Fig. 9: The visualization ablation results of the components.

a Input. b Mask. c w/o SKM. d w/o SKM & MSCA. e Ours. f GT. The figures below (a1–f1) and (a2–f2) are the corresponding partially enlarged details. Mask ratio is 20–30%.

a Input. b Mask. c w/o SKM. d w/o SKM & MSCA. e Ours. f GT. The figures below (a1–f1) and (a2–f2) are the corresponding partially enlarged details. Mask ratio is 20–30%.