def __init__(self, n_class, in_ch, n_e=24*2, n_h=48*2, g_size=32, n_step=8, scale=3, var=7.5):
w = math.sqrt(2) # MSRA scaling
n_ch_after_core = 256
n_emb_l =128
# n_ch_in_core = n_h*2*math.ceil(g_size/8)*math.ceil(g_size/8)
n_ch_in_core = n_h*math.ceil(g_size/4)*math.ceil(g_size/4)
super().__init__(
emb_l = L.Linear(2, n_emb_l), # embed location
emb_x = L.MLPConvolution2D( \
in_ch*scale, (n_e, n_e, n_e), 5, pad=2, wscale=w), # embed image
conv_loc_to_glimpse = L.Linear(n_emb_l, n_ch_in_core), # loc to glimpse. output channel is n_h, if FC.
conv_image_to_glimpse = L.MLPConvolution2D( \
n_e, (n_h, n_h, n_h), 5, pad=2, wscale=w), # image to glimpse
core_lstm = L.LSTM(n_ch_in_core, n_ch_after_core), # core LSTM. in the paper, named recurrent network.
fc_ha = L.Linear(n_ch_after_core, n_class), # core to action(from core_lstm to action)
fc_hl = L.Linear(n_ch_after_core, 2), # core to loc(from core_lstm to loc). in the paper, named emission network.
fc_hb = L.Linear(n_ch_after_core, 1), # core to baseline(from core_lstm to baseline)
)
self.g_size = g_size
self.n_step = n_step
self.scale = scale
self.var = var
self.train = True
self.n_class = n_class
self.active_learn = False
def __init__(self):
w = math.sqrt(2) # MSRA scaling
super(NIN, self).__init__(
mlpconv1=L.MLPConvolution2D(
3, (96, 96, 96), 11, stride=4, wscale=w),
mlpconv2=L.MLPConvolution2D(
96, (256, 256, 256), 5, pad=2, wscale=w),
mlpconv3=L.MLPConvolution2D(
256, (384, 384, 384), 3, pad=1, wscale=w),
mlpconv4=L.MLPConvolution2D(
384, (1024, 1024, 1000), 3, pad=1, wscale=w),
)
self.train = True
def __init__(self, n_class=1000): # 1000 is for ImageNet
super(NIN, self).__init__()
conv_init = I.HeNormal() # MSRA scaling
self.n_class = n_class
with self.init_scope():
self.mlpconv1 = L.MLPConvolution2D(
None, (96, 96, 96), 11, stride=4, conv_init=conv_init)
self.mlpconv2 = L.MLPConvolution2D(
None, (256, 256, 256), 5, pad=2, conv_init=conv_init)
self.mlpconv3 = L.MLPConvolution2D(
None, (384, 384, 384), 3, pad=1, conv_init=conv_init)
self.mlpconv4 = L.MLPConvolution2D(
None, (1024, 1024, self.n_class), 3, pad=1, conv_init=conv_init)
def __init__(self):
w = math.sqrt(2)
super(NIN, self).__init__(
mlpconv1=L.MLPConvolution2D(3, (192, 160, 96), 5, pad=2, wscale=w),
mlpconv2=L.MLPConvolution2D(96, (192, 192, 192),
5,
pad=2,
wscale=w),
mlpconv3=L.MLPConvolution2D(192, (192, 192, 10),
3,
pad=1,
wscale=w),
)
self.train = True
def setUp(self):
self.mlp = links.MLPConvolution2D(3, (96, 96, 96),
11,
activation=functions.sigmoid)
self.mlp.to_gpu()
self.x = cuda.cupy.zeros((10, 3, 20, 20), dtype=numpy.float32)
self.gy = cuda.cupy.zeros((10, 96, 10, 10), dtype=numpy.float32)
def __init__(self, n_class=10):
w = chainer.initializers.HeNormal()
super(NIN, self).__init__()
with self.init_scope():
self.mlpconv1 = L.MLPConvolution2D(3, (192, 160, 96),
5,
pad=2,
conv_init=w)
self.mlpconv2 = L.MLPConvolution2D(96, (192, 192, 192),
5,
pad=2,
conv_init=w)
self.mlpconv3 = L.MLPConvolution2D(192, (192, 192, n_class),
3,
pad=1,
conv_init=w)
def setUp(self):
self.mlp = links.MLPConvolution2D(3, (96, 96, 96),
11,
activation=functions.sigmoid)
with testing.assert_warns(DeprecationWarning):
self.mlp.to_gpu()
self.x = cuda.cupy.zeros((10, 3, 20, 20), dtype=numpy.float32)
self.gy = cuda.cupy.zeros((10, 96, 10, 10), dtype=numpy.float32)
def test_valid_instantiation_ksize_is_not_none(self):
l = links.MLPConvolution2D(
self.in_channels, self.out_channels, self.ksize, self.stride,
self.pad, functions.relu, conv_init=None, bias_init=None)
self.assertEqual(len(l), 2)
self.assertEqual(l[0].W.shape,
(self.out_channels[0], self.in_channels,
self.ksize, self.ksize))
self.assertEqual(l[1].W.shape,
(self.out_channels[1], self.out_channels[0], 1, 1))
def __init__(self):
conv_init = I.HeNormal() # MSRA scaling
super(NIN, self).__init__(
mlpconv1=L.MLPConvolution2D(None, (96, 96, 96),
11,
stride=4,
conv_init=conv_init),
mlpconv2=L.MLPConvolution2D(None, (256, 256, 256),
5,
pad=2,
conv_init=conv_init),
mlpconv3=L.MLPConvolution2D(None, (384, 384, 384),
3,
pad=1,
conv_init=conv_init),
mlpconv4=L.MLPConvolution2D(None, (1024, 1024, 1000),
3,
pad=1,
conv_init=conv_init),
)
def __init__(self, image_colors, class_labels):
w = math.sqrt(2) # MSRA scaling
super(NIN, self).__init__(
mlpconv1=L.MLPConvolution2D(None, (96, 96, 96),
11,
stride=4,
wscale=w),
mlpconv2=L.MLPConvolution2D(None, (256, 256, 256),
5,
pad=2,
wscale=w),
mlpconv3=L.MLPConvolution2D(None, (384, 384, 384),
3,
pad=1,
wscale=w),
mlpconv4=L.MLPConvolution2D(None, (1024, 1024, class_labels),
3,
pad=1,
wscale=w),
)
self.train = True
def __init__(self, compute_accuracy=False):
super(NIN, self).__init__()
self.compute_accuracy = compute_accuracy
conv_init = I.HeNormal() # MSRA scaling
with self.init_scope():
self.mlpconv1 = L.MLPConvolution2D(None, (96, 96, 96),
11,
stride=4,
conv_init=conv_init)
self.mlpconv2 = L.MLPConvolution2D(None, (256, 256, 256),
5,
pad=2,
conv_init=conv_init)
self.mlpconv3 = L.MLPConvolution2D(None, (384, 384, 384),
3,
pad=1,
conv_init=conv_init)
self.mlpconv4 = L.MLPConvolution2D(None, (1024, 1024, 1000),
3,
pad=1,
conv_init=conv_init)
def test_valid_instantiation_in_channels_is_omitted(self):
l = links.MLPConvolution2D(
self.out_channels, self.ksize, stride=self.stride, pad=self.pad,
activation=functions.relu, conv_init=None, bias_init=None)
x = numpy.random.uniform(
-1, 1, (10, self.in_channels, 10, 10)).astype(numpy.float32)
l(x) # create weight tensors of convolutions by initialization
self.assertEqual(len(l), 2)
self.assertEqual(l[0].W.shape,
(self.out_channels[0], self.in_channels,
self.ksize, self.ksize))
self.assertEqual(l[1].W.shape,
(self.out_channels[1], self.out_channels[0], 1, 1))
def test_forbid_wscale_as_a_keyword_argument(self):
with self.assertRaises(ValueError):
links.MLPConvolution2D(self.in_channels,
self.out_channels,
wscale=1)
def test_forbid_wscale_as_a_positional_argument(self):
with self.assertRaises(TypeError):
# 7th positional argument was wscale in v1
links.MLPConvolution2D(self.in_channels, self.out_channels, None,
self.stride, self.pad, functions.relu, 1)
def __init__(self,
n_dims,
in_channels,
hidden_channels,
out_channel,
kernel_size=3,
initialW=initializers.HeNormal(),
initial_bias=None,
block_type='default',
is_residual=False,
batch_norm=False):
self.n_dims = n_dims
self.in_channels = in_channels
self.hidden_channels = hidden_channels
self.out_channel = out_channel
self.kernel_size = kernel_size
self.initialW = initialW
self.initial_bias = initial_bias
self.block_type = block_type
self.is_residual = is_residual
self.batch_norm = batch_norm
pad = self.kernel_size // 2 if self.kernel_size % 2 == 0 else (
self.kernel_size - 1) // 2
super().__init__()
with self.init_scope():
if self.block_type == 'default':
self.conv_1 = L.ConvolutionND(self.n_dims,
self.in_channels,
self.hidden_channels,
self.kernel_size,
stride=1,
pad=pad,
initialW=self.initialW,
initial_bias=self.initial_bias)
self.conv_2 = L.ConvolutionND(self.n_dims,
self.hidden_channels,
self.out_channel,
self.kernel_size,
stride=1,
pad=pad,
initialW=self.initialW,
initial_bias=self.initial_bias)
elif self.block_type == 'dilated':
assert self.n_dims != 2, 'Currently, dilated convolution is unsupported in 3D.'
self.conv_1 = L.DilatedConvolution2D(
self.in_channels,
self.hidden_channels,
self.kernel_size,
stride=1,
pad=pad,
dilate=1,
initialW=self.initialW,
initial_bias=self.initial_bias)
self.conv_2 = L.DilatedConvolution2D(
self.hidden_channels,
self.out_channel,
self.kernel_size,
stride=1,
pad=pad * 2,
dilate=2,
initialW=self.initialW,
initial_bias=self.initial_bias)
elif self.block_type == 'mlp':
assert self.n_dims != 2, 'Currently, mlp convolution is unsupported in 3D.'
self.conv_1 = L.MLPConvolution2D(self.in_channels,
[self.hidden_channels] * 3,
self.kernel_size,
stride=1,
pad=pad,
conv_init=self.initialW,
bias_init=self.initial_bias)
self.conv_2 = L.MLPConvolution2D(self.hidden_channels,
[self.out_channel] * 3,
self.kernel_size,
stride=1,
pad=pad,
conv_init=self.initialW,
bias_init=self.initial_bias)
if self.batch_norm:
self.bn_conv_1 = L.BatchNormalization(self.hidden_channels)
self.bn_conv_2 = L.BatchNormalization(self.out_channel)
def setUp(self):
args, kwargs = self.mlpconv_args
self.mlp = links.MLPConvolution2D(*args, **kwargs)
self.x = numpy.zeros((10, 3, 20, 20), dtype=numpy.float32)
def setUp(self):
self.mlp = links.MLPConvolution2D(3, (96, 96, 96),
11,
activation=functions.sigmoid,
use_cudnn=self.use_cudnn)
self.x = numpy.zeros((10, 3, 20, 20), dtype=numpy.float32)
plt.imshow(image)
print(int(400 / 4))
print(int(300 / 4))
print(98 * 4)
print(73 * 4)
#image = image[:,:,:,np.newaxis]
image = np.transpose(image, (2, 0, 1))
image = image[np.newaxis, :, :, :]
print("input image:", image.shape)
#conv1=L.Convolution2D(3, 96, 11)#, stride=)
w = math.sqrt(2) # MSRA scaling
conv1 = L.MLPConvolution2D(3, (96, 96, 96), 11, stride=4, wscale=w)
mlpconv1 = conv1(image)
print("conv result1:", mlpconv1.data.shape)
#rev_image = np.transpose(cv1.data,(1,2,3,0))
#plt.imshow(rev_image[0])
type(mlpconv1)
print("conv result1 image shape:", mlpconv1.data[0].shape)
print("mlpconv1 is")
plot(mlpconv1.data, mlpconv1.data.shape[1])
relu1 = F.relu(mlpconv1)
print("relu result1:", relu1.data.shape)
