Skip to main content
Fig. 8 | Plant Methods

Fig. 8

From: HairNet: a deep learning model to score leaf hairiness, a key phenotype for cotton fibre yield, value and insect resistance

Fig. 8

Effect of leaf selection (L3 and L4) (A), year to year variation (Y1 and Y2) (B) and environment (C) on the model accuracy. The best models (ResNet34 feature extractor with Random Rotation (RR) data augmentation and Adam optimizer) performance on the whole dataset is plotted in each row as reference. Each bar in the figure represents the average accuracy and the range clips on each bar represent the maximum and minimum accuracy of five runs of the same experiment with the best model. The training dataset size of each experiment is plotted on the right y axis of each row of bar plots

Back to article page