Figure 1

Stepwise change in loss and accuracy in the training process. Loss denotes the value of cross entropy at each step. The process of training with images of square-shaped wells is assessed by reference to changes in training loss (a), training accuracy (b), validation loss (c), and validation accuracy (d) with each iteration step. The process of training with images of round-shaped wells is assessed by reference to changes in training loss (e), training accuracy (f), validation loss (g), and validation accuracy (h) with each iteration step.