Fig. 5From: Detecting central fixation by means of artificial neural networks in a pediatric vision screener using retinal birefringence scanningThe combined confusion matrix for the three kinds of data (train, validate, test) from the calibration set. The green diagonal cells show the number and percentage of correct classifications by the trained network. The red diagonal represents the incorrect classifications, which are 0 for each target class and for each output class. The lower right blue square illustrates the overall accuracyBack to article page