def __init__(self, net_file=None, meta_file=None):
"""Initializes DecafNet.
Input:
net_file: the trained network file.
meta_file: the meta information for images.
"""
logging.info('Initializing decafnet...')
try:
if not net_file:
# use the internal decafnet file.
net_file = _JEFFNET_FILE
if not meta_file:
# use the internal meta file.
meta_file = _META_FILE
cuda_decafnet = pickle.load(open(net_file))
meta = pickle.load(open(meta_file))
except IOError:
raise RuntimeError('Cannot find DecafNet files.')
# First, translate the network
self._net = translator.translate_cuda_network(
cuda_decafnet, {'data': (INPUT_DIM, INPUT_DIM, 3)})
# Then, get the labels and image means.
self.label_names = meta['label_names']
self._data_mean = translator.img_cudaconv_to_decaf(
meta['data_mean'], 256, 3)
logging.info('Jeffnet initialized.')
return
def __init__(self, net_file=None, meta_file=None):
"""Initializes DecafNet.
Input:
net_file: the trained network file.(normally,['model_state']['layers'])
meta_file: the meta information for images.
"""
logging.info('Initializing decafnet...')
try:
if not net_file:
# use the internal decafnet file.
net_file = _KITNET_FILE
if not meta_file:
# use the internal meta file.
meta_file = _META_FILE
cuda_decafnet = pickle.load(open(net_file,'rb'))['model_state']['layers']
meta = pickle.load(open(meta_file,'rb'))
except IOError:
raise RuntimeError('Cannot find DecafNet files.')
# First, translate the network
self._net = translator.translate_cuda_network(
cuda_decafnet, {'data': (INPUT_DIM, 1)})
# Then, get the labels and image means.
self.label_names = meta['label_names']
self._data_mean = meta['data_mean']
logging.info('Kitnet initialized.')
return
def load_net(fname):
cudaconv_net = IGPUModel.load_checkpoint(fname)
layers = cudaconv_net["model_state"]["layers"]
#Note, data dimensions are hardcoded here - not sure we have that info in the cudaconv object?
decafnet = translator.translate_cuda_network(layers, {'data': (32, 32, 3)})
return decafnet
def __init__(self, net_file=None, meta_file=None):
"""Initializes DecafNet.
Input:
net_file: the trained network file.
meta_file: the meta information for images.
"""
logging.info('Initializing decafnet...')
try:
if not net_file:
# use the internal decafnet file.
net_file = _JEFFNET_FILE
if not meta_file:
# use the internal meta file.
meta_file = _META_FILE
cuda_decafnet = pickle.load(open(net_file))
meta = pickle.load(open(meta_file))
except IOError:
raise RuntimeError('Cannot find DecafNet files.')
# First, translate the network
self._net = translator.translate_cuda_network(
cuda_decafnet, {'data': (INPUT_DIM, INPUT_DIM, 3)})
# Then, get the labels and image means.
self.label_names = meta['label_names']
self._data_mean = translator.img_cudaconv_to_decaf(
meta['data_mean'], 256, 3)
logging.info('Jeffnet initialized.')
return
def load_net(fname):
cudaconv_net = IGPUModel.load_checkpoint(fname)
layers = cudaconv_net["model_state"]["layers"]
#Note, data dimensions are hardcoded here - not sure we have that info in the cudaconv object?
decafnet = translator.translate_cuda_network(layers, {'data': (32, 32, 3)})
return decafnet
def load_from_convnet(filename):
cudaconv_net = IGPUModel.load_checkpoint(filename)
layers = cudaconv_net["model_state"]["layers"]
data_layer = [l for l in layers if l["name"] == "data"][0]
data_consumer = [l for l in layers if data_layer in l.get("inputLayers", [])][0]
input_dim = int(sqrt(data_consumer["imgPixels"][0])) # The width/height of the square images
input_channels = data_consumer["channels"][0] # The number of channels in the input images
return translator.translate_cuda_network(layers, {"data": (input_dim, input_dim, input_channels)})
def setUp(self):
self._layers = pickle.load(
open(os.path.join(_TEST_DATA_DIR, 'layers.pickle')))
self._data = pickle.load(
open(os.path.join(_TEST_DATA_DIR, 'data', 'data_batch_5')))
self._decaf_data = translator.imgs_cudaconv_to_decaf(
self._data['data'][:_BATCH_SIZE], 32, 3)
#self._decaf_labels = self._data['labels'].flatten()[:_BATCH_SIZE]
#self._decaf_labels = self._decaf_labels.astype(np.int)
self._output_shapes = {'data': (32, 32, 3), 'labels': -1}
self._net = translator.translate_cuda_network(self._layers,
self._output_shapes)
self._net.predict(data=self._decaf_data)
def load_from_convnet(filename):
cudaconv_net = IGPUModel.load_checkpoint(filename)
layers = cudaconv_net["model_state"]["layers"]
data_layer = [l for l in layers if l['name'] == 'data'][0]
data_consumer = [
l for l in layers if data_layer in l.get('inputLayers', [])
][0]
input_dim = int(sqrt(data_consumer['imgPixels']
[0])) # The width/height of the square images
input_channels = data_consumer['channels'][
0] # The number of channels in the input images
return translator.translate_cuda_network(
layers, {'data': (input_dim, input_dim, input_channels)})
def setUp(self):
self._layers = pickle.load(open(os.path.join(_TEST_DATA_DIR,
'layers.pickle')))
self._data = pickle.load(open(os.path.join(_TEST_DATA_DIR,
'data',
'data_batch_5')))
self._decaf_data = translator.imgs_cudaconv_to_decaf(
self._data['data'][:_BATCH_SIZE], 32, 3)
#self._decaf_labels = self._data['labels'].flatten()[:_BATCH_SIZE]
#self._decaf_labels = self._decaf_labels.astype(np.int)
self._output_shapes = {'data': (32, 32, 3), 'labels': -1}
self._net = translator.translate_cuda_network(
self._layers, self._output_shapes)
self._net.predict(data=self._decaf_data)