def test_bottleneck_block_with_superpixel_layer(self):
"""Test for creating a model with fused bottleneck block arguments."""
images = tf.zeros((10, 128, 128, 3), dtype=tf.float32)
global_params = efficientnet_model.GlobalParams(
1.0,
1.0,
0,
'channels_last',
num_classes=10,
batch_norm=utils.TpuBatchNormalization)
blocks_args = [
efficientnet_model.BlockArgs(
kernel_size=3,
num_repeat=3,
input_filters=3,
output_filters=6,
expand_ratio=6,
id_skip=True,
strides=[2, 2],
conv_type=0,
fused_conv=0,
super_pixel=1)
]
model = efficientnet_model.Model(blocks_args, global_params)
outputs = model(images, training=True)
self.assertEqual((10, 10), outputs.shape)
def test_variables(self):
"""Test for variables in blocks to be included in `model.variables`."""
images = tf.zeros((10, 128, 128, 3), dtype=tf.float32)
global_params = efficientnet_model.GlobalParams(
1.0,
1.0,
0,
'channels_last',
num_classes=10,
batch_norm=utils.TpuBatchNormalization)
blocks_args = [
efficientnet_model.BlockArgs(
kernel_size=3,
num_repeat=3,
input_filters=3,
output_filters=6,
expand_ratio=6,
id_skip=False,
strides=[2, 2],
se_ratio=0.8,
conv_type=0,
fused_conv=0,
super_pixel=0)
]
model = efficientnet_model.Model(blocks_args, global_params)
_ = model(images, training=True)
var_names = {var.name for var in model.variables}
self.assertIn('model/blocks_0/conv2d/kernel:0', var_names)
def test_reduction_endpoint_with_single_block_without_sp(self):
"""Test reduction point with single block/layer."""
images = tf.zeros((10, 128, 128, 3), dtype=tf.float32)
global_params = efficientnet_model.GlobalParams(
1.0,
1.0,
0,
'channels_last',
num_classes=10,
batch_norm=utils.TpuBatchNormalization)
blocks_args = [
efficientnet_model.BlockArgs(
kernel_size=3,
num_repeat=1,
input_filters=3,
output_filters=6,
expand_ratio=6,
id_skip=False,
strides=[2, 2],
se_ratio=0.8,
conv_type=0,
fused_conv=0,
super_pixel=0)
]
model = efficientnet_model.Model(blocks_args, global_params)
_ = model(images, training=True)
self.assertIn('reduction_1', model.endpoints)
# single block should have one and only one reduction endpoint
self.assertNotIn('reduction_2', model.endpoints)
def efficientnet_lite(width_coefficient=None,
depth_coefficient=None,
dropout_rate=0.2,
survival_prob=0.8):
"""Creates a efficientnet model."""
global_params = efficientnet_model.GlobalParams(
blocks_args=_DEFAULT_BLOCKS_ARGS,
batch_norm_momentum=0.99,
batch_norm_epsilon=1e-3,
dropout_rate=dropout_rate,
survival_prob=survival_prob,
data_format='channels_last',
num_classes=1000,
width_coefficient=width_coefficient,
depth_coefficient=depth_coefficient,
depth_divisor=8,
min_depth=None,
relu_fn=tf.nn.relu6, # Relu6 is for easier quantization.
# The default is TPU-specific batch norm.
# The alternative is tf.layers.BatchNormalization.
batch_norm=utils.TpuBatchNormalization, # TPU-specific requirement.
clip_projection_output=False,
fix_head_stem=True, # Don't scale stem and head.
local_pooling=True, # special cases for tflite issues.
use_se=False) # SE is not well supported on many lite devices.
return global_params
def efficientnet(width_coefficient=None,
depth_coefficient=None,
dropout_rate=0.2,
survival_prob=0.8):
"""Creates a efficientnet model."""
global_params = efficientnet_model.GlobalParams(
blocks_args=_DEFAULT_BLOCKS_ARGS,
batch_norm_momentum=0.99,
batch_norm_epsilon=1e-3,
dropout_rate=dropout_rate,
survival_prob=survival_prob,
data_format='channels_last',
num_classes=1000,
width_coefficient=width_coefficient,
depth_coefficient=depth_coefficient,
depth_divisor=8,
min_depth=None,
relu_fn=tf.nn.swish,
# The default is TPU-specific batch norm.
# The alternative is tf.layers.BatchNormalization.
batch_norm=utils.TpuBatchNormalization, # TPU-specific requirement.
use_se=True,
clip_projection_output=False)
return global_params
