keras中的中间层的输出通过DirectoryIterator作为输入

Question

Is there a way to get the output of a intermediate layer in Keras passing a DirectoryIterator as input?

I've already read the documentation of Keras related to this and this question but as far as I undertood, those approaches use single samples as input.

I want to do this because it is more simple than read sample by sample and less computationally expensive than put everything on memory.

I'm using TensorFlow as backend.

Answer 1

desiredOutputs = model.get_layer('layer_name').outputs #or model.layers[n].outputs
partialModel = Model(model.inputs,desiredOutputs)

Then use partialModel to fit or predict with the directory iterator.