def _build_network(self):
import config
if config.model_type == MODEL_TYPE_vgg16:
from nets import vgg
with slim.arg_scope(
[slim.conv2d],
activation_fn=tf.nn.relu,
weights_regularizer=slim.l2_regularizer(
config.weight_decay),
weights_initializer=tf.contrib.layers.xavier_initializer(),
biases_initializer=tf.zeros_initializer()):
with slim.arg_scope([slim.conv2d, slim.max_pool2d],
padding='SAME') as sc:
self.arg_scope = sc
self.net, self.end_points = vgg.basenet(inputs=self.inputs,
pooling='MAX')
elif config.model_type == MODEL_TYPE_vgg16_no_dilation:
from nets import vgg
with slim.arg_scope(
[slim.conv2d],
activation_fn=tf.nn.relu,
weights_regularizer=slim.l2_regularizer(
config.weight_decay),
weights_initializer=tf.contrib.layers.xavier_initializer(),
biases_initializer=tf.zeros_initializer()):
with slim.arg_scope([slim.conv2d, slim.max_pool2d],
padding='SAME') as sc:
self.arg_scope = sc
self.net, self.end_points = vgg.basenet(inputs=self.inputs,
dilation=False,
pooling='MAX')
else:
raise ValueError('model_type not supported:%s' %
(config.model_type))
def model_vgg(images, weight_decay=1e-5, is_training=True):
'''
define the model, we use slim's implemention of resnet
'''
images = mean_image_subtraction(images)
with slim.arg_scope(resnet_v1.resnet_arg_scope(weight_decay=weight_decay)):
logits, end_points = vgg.basenet(images, scope='vgg_16')
with tf.variable_scope('feature_fusion', values=[end_points.values]):
batch_norm_params = {
'decay': 0.997,
'epsilon': 1e-5,
'scale': True,
'is_training': is_training
}
with slim.arg_scope(
[slim.conv2d],
activation_fn=tf.nn.relu,
normalizer_fn=slim.batch_norm,
normalizer_params=batch_norm_params,
weights_regularizer=slim.l2_regularizer(weight_decay)):
feature_maps = [
end_points['fc7'], end_points['conv5_3'],
end_points['conv4_3'], end_points['conv3_3'],
end_points['conv2_2']
]
pixel_1 = slim.conv2d(end_points['fc7'], 2, 1) + slim.conv2d(
end_points['conv5_3'], 2, 1)
pixel_2 = unpool(pixel_1) + slim.conv2d(end_points['conv4_3'], 2,
1)
pixel_3 = unpool(pixel_2) + slim.conv2d(end_points['conv3_3'], 2,
1)
pixel_cls = slim.conv2d(pixel_3, 2, 1)
print('pixel_shape:{}'.format(pixel_cls.shape))
link_1 = slim.conv2d(end_points['fc7'], 16, 1) + slim.conv2d(
end_points['conv5_3'], 16, 1)
link_2 = unpool(link_1) + slim.conv2d(end_points['conv4_3'], 16, 1)
link_3 = unpool(link_2) + slim.conv2d(end_points['conv3_3'], 16, 1)
link_cls = slim.conv2d(link_3, 16, 1)
print('link_shape:{}'.format(link_cls.shape))
return pixel_cls, link_cls