using googleNet instead of caffeNet

[Zahra Aminolroaya]

I used this link to test my images with pretrained models. I want to  use googleNet instead of caffeNet.  here is my code:

res=np.empty([len(id),10]);
plt.rcParams['figure.figsize'] = (10, 10)
plt.rcParams['image.interpolation'] = 'nearest'
plt.rcParams['image.cmap'] = 'gray'
caffe.set_mode_cpu()
net = caffe.Net('./models/bvlc_googlenet/deploy.prototxt',
               './models/bvlc_googlenet/bvlc_googlenet.caffemodel',
                caffe.TEST)
transformer = caffe.io.Transformer({'data': net.blobs['data'].data.shape})
transformer.set_transpose('data', (2, 0, 1))
transformer.set_mean('data', np.load('./python/caffe/imagenet/ilsvrc_2012_mean.npy').mean(1).mean(1))
transformer.set_raw_scale('data', 255)
transformer.set_channel_swap('data', (2, 1, 0))
net.blobs['data'].reshape(1, 3, 224, 224);
imagenet_labels_filename = './data/ilsvrc12/synset_words.txt'
labels = np.loadtxt(imagenet_labels_filename, str, delimiter='\t')

I just changed two things : 

     1- first the net direction ((net = caffe.Net('./models/bvlc_googlenet/deploy.prototxt',

               './models/bvlc_googlenet/bvlc_googlenet.caffemodel',
                caffe.TEST) ))

   2-second the reshape to 224 * 224 ((net.blobs['data'].reshape(1, 3, 224, 224)))

I want to know that if it is enough. because I got the results as same as caffenet !!

shall I change this : transformer.set_mean('data', np.load('./python/caffe/imagenet/ilsvrc_2012_mean.npy').mean(1).mean(1))??

Because I know that googleNet worked on ilsvrc2014!!!

thanks