What is the correct way of using MemoryData layer using pycaffe?

[Frank Darius]

I'm planning to do real-time augmentation in `caffe`. and these are the steps I have taken so far:   

1.Replace `Data` layer with `MemoryData` in the network:

    name: "test_network"
    layer {
      name: "test_data"
      type: "MemoryData"
      top: "data"
      top: "label"
      include {
        phase: TRAIN
      }
      memory_data_param {
       batch_size: 32
       channels: 3
       height: 32
       width: 32
      }
   
    }
    layer {
      name: "test_data"
      type: "MemoryData"
      top: "data"
      top: "label"
      include {
        phase: TEST
      }
       memory_data_param {
       batch_size: 32
       channels: 3
       height: 32
       width: 32
      }
    }

and this is the code for training : 

    caffe.set_mode_gpu()
    maxIter = 100
    batch_size = 32
    j = 0
    for i in range(maxIter):
        #fetch images
        batch = seq.augment_images(np.transpose(data_train[j: j+batch_size],(0,2,3,1)))
        print('batch-{0}-{1}'.format(j,j+batch_size))
        #set input and solve
        batch = batch.reshape(-1,3,32,32).astype(np.float32)
        net.set_input_arrays(batch, label_train[j: j+batch_size].astype(np.float32))
        j = j + batch_size + 1
        solver.step(1)

but when the code reaches to the net.set_input_arrays(), it crashes with this error:

   

 W0405 20:53:19.679730  4640 memory_data_layer.cpp:90] MemoryData does not transform array data on Reset()
    I0405 20:53:19.713727  4640 solver.cpp:337] Iteration 0, Testing net (#0)
    I0405 20:53:19.719229  4640 net.cpp:685] Ignoring source layer accuracy_training
    F0405 20:53:19.719229  4640 memory_data_layer.cpp:110] Check failed: data_ MemoryDataLayer needs to be initalized by calling Reset
    *** Check failure stack trace: ***

I cant find the reset() method anywhere, Is this layer even supported properly? there is zero documentation about this. 

The c++ interface has some explanations! but the python wrapper has nothing! 

At the very least an example for this layer is appreciated.