Figure 7.

Visualization of top-k CutMix food image augmentation with $k=0,1,3,5,10$. The original image is from the stewed beef food type and the augmented foods are from steak, mixed pasta dishes and burgers, which were selected based on visual similarity.