Table 6 Comparison of the effect of various data augmentation techniques on the performance of the top two module pairs. Data augmentation methods tested here are: image resizing (RS) to 512 x 512 px (RS512), 768 x 768 px (RS768), 1024 x 1024 px (RS1024); random cropping (RC) to 1536 x 1536 px; random flip (RF). The best performing model + data augmentation pair is highlighted in grey value

From: HairNet2: deep learning to quantify cotton leaf hairiness, a complex genetic and environmental trait