Noiredd

import numpy as np

def load_results():
  path = '' #if you run from a different folder
  a = np.load(path+'test_forward_start_end_ipblob.npy')
  b = np.load(path+'test_forward_start_end_manual.npy')
  c = np.load(path+'test_backward_start_end_convblob.npy')
  d = np.load(path+'test_backward_start_end_manual.npy')
  return a, b, c, d

Noiredd

#Replace the corresponding functions in test_net.py
    def test_forward_start_end(self):
        conv_blob=self.net.blobs['conv'];
        ip_blob=self.net.blobs['ip_blob'];
        sample_data=np.random.uniform(size=conv_blob.data.shape);
        sample_data=sample_data.astype(np.float32);
        #"""Uncomment the following to load previously stored initializations"""
        #sample_data=np.load('test_forward_start_end_data.npy');
        #self.net.params['ip'][0].data[...]=np.load('test_forward_start_end_weights.npy');
        #self.net.params['ip'][1].data[...]=np.load('test_forward_start_end_biases.npy');

Noiredd

from skimage.transform import rescale
import numpy as np
import caffe

inputArray = np.array( [[0.1, 0.2, 0.3, 0.4], [0.4, 0.3, 0.2, 0.1]] )
scikitResult = rescale(inputArray, scale=2, mode='constant', cval=0)

net = caffe.Net('5173_vsScikit.pt', caffe.TRAIN)
net.blobs['data'].data[0][0][...] = inputArray
net.forward()

Noiredd

input: "data"
input_shape {
  dim: 1
  dim: 1
  dim: 2
  dim: 4
}
layer {
  name: "x2"
  type: "Deconvolution"

Noiredd

import caffe
n = caffe.Net('5173.pt', caffe.TRAIN)
for l in n.layers[2:]:
	print l.blobs[0].data
	print 'These should be equal:', l.blobs[0].data.flatten()[0], l.blobs[0].data.flatten()[-1]

Noiredd

input: "data"
input_shape {
  dim: 1
  dim: 1
  dim: 350
  dim: 576
}
layer {
  name: "bilinear3"
  type: "Deconvolution"

Noiredd

input: "data"
input_shape {
  dim: 1
  dim: 3
  dim: 227
  dim: 227
}
layer {
  name: "conv1"
  type: "Convolution"