def visualize(self, layout_dir='LR', display=False):
"""
Create and optionally display an image of the net structure.
:param layout_dir: string in ['LR', 'TB', 'BT'].
Short string for graph layout direction.
:param display: bool.
If set to ``True``, displays the graphic in a window. Press enter
to close it.
:returns: 3D numpy array.
Graphic of the visualization as (H, W, C) image in BGR format.
"""
if _draw is None or _cv2 is None: # pragma: no cover
raise Exception('Drawing is not available!')
else: # pragma: no cover
with _NamedTemporaryFile(mode='w+b', suffix='.png') as tmpfile:
_draw.draw_net_to_file(self.to_pbuf_message(),
tmpfile.name,
rankdir=layout_dir)
result_image = _cv2.imread(tmpfile.name)
assert result_image is not None
if display: # pragma: no cover
_cv2.imshow(self.name, result_image)
_cv2.waitKey(0)
_cv2.destroyWindow(self.name)
return result_image
def visualize(self,
layout_dir='LR',
display=False):
"""
Create and optionally display an image of the net structure.
:param layout_dir: string in ['LR', 'TB', 'BT'].
Short string for graph layout direction.
:param display: bool.
If set to ``True``, displays the graphic in a window. Press enter
to close it.
:returns: 3D numpy array.
Graphic of the visualization as (H, W, C) image in BGR format.
"""
if _draw is None or _cv2 is None:
raise Exception('Drawing is not available!')
with _NamedTemporaryFile(mode='w+b', suffix='.png') as tmpfile:
_draw.draw_net_to_file(self.to_pbuf_message(),
tmpfile.name,
rankdir=layout_dir)
result_image = _cv2.imread(tmpfile.name)
assert result_image is not None
if display:
_cv2.imshow(self.name, result_image)
_cv2.waitKey(0)
_cv2.destroyWindow(self.name)
return result_image
def manipulate_train(ori, target_train, **kwargs):
train_pb = caffe_pb2.NetParameter()
if cfg.MODEL.DIFFERENT_DILATION.ENABLE:
ori = 'models/train_different_dilation_template.prototxt'
with open(ori, 'r') as f:
train_txt = f.read()
txtf.Merge(train_txt, train_pb)
train_pb = _add_dimension_reduction(train_pb)
train_pb = _apply_mult_lr(train_pb)
train_vis_file = '.'.join(target_train.rsplit('.')[:-1]) + '.jpg'
with open(target_train, 'w') as f:
f.write(str(train_pb))
draw_net_to_file(train_pb, train_vis_file, 'LR', 0)
tb.sess.add_image('train_net', train_vis_file, wall_time=0, step=0)
return None
with open(ori, 'r') as f:
train_txt = f.read()
txtf.Merge(train_txt, train_pb)
train_pb = _add_dimension_reduction(train_pb)
train_pb = _apply_mult_lr(train_pb)
train_vis_file = '.'.join(target_train.rsplit('.')[:-1]) + '.jpg'
with open(target_train, 'w') as f:
f.write(str(train_pb))
draw_net_to_file(train_pb, train_vis_file, 'LR', 0)
tb.sess.add_image('train_net', train_vis_file, wall_time=0, step=0)
def print_network(prototxt_filename):
'''
Draw the ANN architecture
'''
_net = caffe.proto.caffe_pb2.NetParameter()
f = open(prototxt_filename)
google.protobuf.text_format.Merge(f.read(), _net)
draw.draw_net_to_file(_net, prototxt_filename + '.png')
print('Draw ANN done!')
def draw_from_proto(fname, wname):
if os.path.isdir(fname):
sh.mkdir("-p", wname)
for base in os.listdir(fname):
sub_fname = os.path.join(fname, base)
if os.path.isdir(sub_fname):
sub_wname = base
else:
sub_wname = base + ".png"
draw_from_proto(sub_fname, os.path.join(wname, sub_wname))
else:
if not (fname.endswith(".pt") or fname.endswith(".prototxt")):
print "Ignore file %s according to its suffix, not `pt` or `prototxt`" % fname
return
if "solver" in fname:
print "Regard file %s as a solver prototxt, ignore!" % fname
return
net_param = caffe_pb2.NetParameter()
text_format.Merge(open(fname, "r").read(), net_param)
draw.draw_net_to_file(net_param, wname)
def manipulate_test(ori, target_test, **kwargs): # TODO more elegant editing
test_pb = caffe_pb2.NetParameter()
if cfg.MODEL.DIFFERENT_DILATION.ENABLE:
ori = 'models/test_different_dilation_template.prototxt'
with open(ori, 'r') as f:
test_txt = f.read()
txtf.Merge(test_txt, test_pb)
test_pb = _add_dimension_reduction(test_pb)
test_vis_file = '.'.join(target_test.rsplit('.')[:-1]) + '.jpg'
with open(target_test, 'w') as f:
f.write(str(test_pb))
draw_net_to_file(test_pb, test_vis_file, 'LR', 1)
tb.sess.add_image('test_net', test_vis_file, wall_time=0, step=0)
return None
with open(ori, 'r') as f:
test_txt = f.read()
txtf.Merge(test_txt, test_pb)
test_pb = _add_dimension_reduction(test_pb)
test_vis_file = '.'.join(target_test.rsplit('.')[:-1]) + '.jpg'
with open(target_test, 'w') as f:
f.write(str(test_pb))
draw_net_to_file(test_pb, test_vis_file, 'LR', 1)
tb.sess.add_image('test_net', test_vis_file, wall_time=0, step=0)
def save_cnn_graph(path_model, name_model, phase, phase_name):
net_parameter = caffe_pb2.NetParameter()
text_format.Merge(open(path_model).read(), net_parameter)
draw_net_to_file(net_parameter, const.PATH_TO_IMAGE_DIR + "/network_graph_" + name_model + "_" + phase_name + ".png", "BT", phase)
