Skip to main content

Table 7 Results comparing augmentations [36]

From: A survey on Image Data Augmentation for Deep Learning

Quantitative results on dogs vs. goldfish
Dogs vs goldfish
Augmentation Val. acc.
None 0.855
Traditional 0.890
GANs 0.865
Neural + no loss 0.915
Neural + content loss 0.900
Neural + style 0.890
Control 0.840
Quantitative results on dogs vs cats
Dogs vs cat
Augmentation Val. acc.
None 0.705
Traditional 0.775
GANs 0.720
Neural + no loss 0.765
Neural + content loss 0.770
Neural + style 0.740
Control 0.710
MNIST 0′s and 8′s
Augmentation Val. acc.
None 0.972
Neural + no loss 0.975
Neural + content loss 0.968