Appreciate your work for RS CD. I have two questions:
In this paper, do the comparison methods apply the data augmentations? Can you illustrate the data augmentations about shift and scale in detail?
In this code, I want to train my own dataset by your model. I saw your code receives [batch_size, height, width, 6] as input, and [batch_size, height, width, 1] as output.
However, when I set my dataloader
You have compile model in the Nest_Net2 model, so I can using train_on_batch, right? And what kind of input& output shape should I feed into the model?
Appreciate your work for RS CD. I have two questions:
In this paper, do the comparison methods apply the data augmentations? Can you illustrate the data augmentations about shift and scale in detail?
In this code, I want to train my own dataset by your model. I saw your code receives [batch_size, height, width, 6] as input, and [batch_size, height, width, 1] as output. However, when I set my dataloader
model = Nest_Net2(input_shape=[256,256,6], deep_supervision= False) x_train, y_train = dataloader.next() // x_train.shape = [8, 256, 256, 6], y_train.shape = [8, 256, 256, 1] model.train_on_batch(x_train, y_train) //error
You have compile model in the Nest_Net2 model, so I can using train_on_batch, right? And what kind of input& output shape should I feed into the model?