What does mean means when creating a net in caffe?

[X.T. Li]

I'm a novice to caffe now following caffe tutorial on ipython notebook: http://nbviewer.ipython.org/github/BVLC/caffe/blob/master/examples/classification.ipynb
I noticed these codes when creating a net:

net = caffe.Classifier(MODEL_FILE, PRETRAINED,
                       mean=np.load(caffe_root + 'python/caffe/imagenet/ilsvrc_2012_mean.npy').mean(1).mean(1),
                       channel_swap=(2,1,0),
                       raw_scale=255,
                       image_dims=(256, 256))

...

However, I don't understand what the second line

mean=np.load(caffe_root + 'python/caffe/imagenet/ilsvrc_2012_mean.npy').mean(1).mean(1)
means? 
Also, where can I find documentations about this Classifier function? I don't see it on the API documentation.
Could anybody help me？