TABLE 1.

Hyperparameter settings used for model training.

Hyperparameter	Value
Batch size per GPU	2
Epochs	30
Iteration of each epoch	30
Optimizer	SGD
Learning rate for encoder	0.02
Learning rate for decoder	0.02
Power in poly to drop learning rate	0.9
Momentum for SGD	0.9
Weights regularizer	0.0001
Weighting of deep supervision loss	0.4

Abrreviations: GPU, graphics processing unit; SGD, stochastic gradient descent.