def test_resnet_creation(self, model_id):
"""Test creation of ResNet models."""
network = backbones.ResNet(model_id=model_id,
se_ratio=0.0,
norm_momentum=0.99,
norm_epsilon=1e-5)
backbone_config = backbones_cfg.Backbone(type='resnet',
resnet=backbones_cfg.ResNet(
model_id=model_id,
se_ratio=0.0))
norm_activation_config = common_cfg.NormActivation(norm_momentum=0.99,
norm_epsilon=1e-5,
use_sync_bn=False)
factory_network = factory.build_backbone(
input_specs=tf.keras.layers.InputSpec(shape=[None, None, None, 3]),
backbone_config=backbone_config,
norm_activation_config=norm_activation_config)
network_config = network.get_config()
factory_network_config = factory_network.get_config()
self.assertEqual(network_config, factory_network_config)
def test_model_creation(self, project_dim, num_proj_layers, ft_proj_idx):
input_size = 224
inputs = np.random.rand(2, input_size, input_size, 3)
input_specs = tf.keras.layers.InputSpec(
shape=[None, input_size, input_size, 3])
tf.keras.backend.set_image_data_format('channels_last')
backbone = backbones.ResNet(model_id=50, activation='relu',
input_specs=input_specs)
projection_head = simclr_head.ProjectionHead(
proj_output_dim=project_dim,
num_proj_layers=num_proj_layers,
ft_proj_idx=ft_proj_idx
)
num_classes = 10
supervised_head = simclr_head.ClassificationHead(
num_classes=10
)
model = simclr_model.SimCLRModel(
input_specs=input_specs,
backbone=backbone,
projection_head=projection_head,
supervised_head=supervised_head,
mode=simclr_model.PRETRAIN
)
outputs = model(inputs)
projection_outputs = outputs[simclr_model.PROJECTION_OUTPUT_KEY]
supervised_outputs = outputs[simclr_model.SUPERVISED_OUTPUT_KEY]
self.assertAllEqual(projection_outputs.shape.as_list(),
[2, project_dim])
self.assertAllEqual([2, num_classes],
supervised_outputs.numpy().shape)
def test_serialize_deserialize(self, level, low_level, shared_decoder):
"""Validate the network can be serialized and deserialized."""
num_classes = 10
backbone = backbones.ResNet(model_id=50)
semantic_decoder = aspp.ASPP(level=level, dilation_rates=[6, 12, 18])
if shared_decoder:
instance_decoder = semantic_decoder
else:
instance_decoder = aspp.ASPP(level=level,
dilation_rates=[6, 12, 18])
semantic_head = panoptic_deeplab_heads.SemanticHead(
num_classes,
level=level,
low_level=low_level,
low_level_num_filters=(64, 32))
instance_head = panoptic_deeplab_heads.InstanceHead(
level=level, low_level=low_level, low_level_num_filters=(64, 32))
post_processor = panoptic_deeplab_merge.PostProcessor(
output_size=[640, 640],
center_score_threshold=0.1,
thing_class_ids=[1, 2, 3, 4],
label_divisor=[256],
stuff_area_limit=4096,
ignore_label=0,
nms_kernel=41,
keep_k_centers=41,
rescale_predictions=True)
model = panoptic_deeplab_model.PanopticDeeplabModel(
backbone=backbone,
semantic_decoder=semantic_decoder,
instance_decoder=instance_decoder,
semantic_head=semantic_head,
instance_head=instance_head,
post_processor=post_processor)
config = model.get_config()
new_model = panoptic_deeplab_model.PanopticDeeplabModel.from_config(
config)
# Validate that the config can be forced to JSON.
_ = new_model.to_json()
# If the serialization was successful, the new config should match the old.
self.assertAllEqual(model.get_config(), new_model.get_config())
def test_sync_bn_multiple_devices(self, strategy, use_sync_bn):
"""Test for sync bn on TPU and GPU devices."""
inputs = np.random.rand(64, 128, 128, 3)
tf.keras.backend.set_image_data_format('channels_last')
with strategy.scope():
backbone = backbones.ResNet(model_id=50, use_sync_bn=use_sync_bn)
model = classification_model.ClassificationModel(
backbone=backbone,
num_classes=1000,
dropout_rate=0.2,
)
_ = model(inputs)
def test_serialize_deserialize(self):
"""Validate the classification net can be serialized and deserialized."""
tf.keras.backend.set_image_data_format('channels_last')
backbone = backbones.ResNet(model_id=50)
model = classification_model.ClassificationModel(
backbone=backbone, num_classes=1000)
config = model.get_config()
new_model = classification_model.ClassificationModel.from_config(config)
# Validate that the config can be forced to JSON.
_ = new_model.to_json()
# If the serialization was successful, the new config should match the old.
self.assertAllEqual(model.get_config(), new_model.get_config())
def test_data_format_gpu(self, strategy, data_format, input_dim):
"""Test for different data formats on GPU devices."""
if data_format == 'channels_last':
inputs = np.random.rand(2, 128, 128, input_dim)
else:
inputs = np.random.rand(2, input_dim, 128, 128)
input_specs = tf.keras.layers.InputSpec(shape=inputs.shape)
tf.keras.backend.set_image_data_format(data_format)
with strategy.scope():
backbone = backbones.ResNet(model_id=50, input_specs=input_specs)
model = classification_model.ClassificationModel(
backbone=backbone,
num_classes=1000,
input_specs=input_specs,
)
_ = model(inputs)
def test_resnet_network_creation(
self, input_size, resnet_model_id, activation):
"""Test for creation of a ResNet-50 classifier."""
inputs = np.random.rand(2, input_size, input_size, 3)
tf.keras.backend.set_image_data_format('channels_last')
backbone = backbones.ResNet(
model_id=resnet_model_id, activation=activation)
self.assertEqual(backbone.count_params(), 23561152)
num_classes = 1000
model = classification_model.ClassificationModel(
backbone=backbone,
num_classes=num_classes,
dropout_rate=0.2,
)
self.assertEqual(model.count_params(), 25610152)
logits = model(inputs)
self.assertAllEqual([2, num_classes], logits.numpy().shape)
def test_serialize_deserialize(self):
"""Validate the network can be serialized and deserialized."""
num_classes = 3
backbone = backbones.ResNet(model_id=50)
decoder = fpn.FPN(
input_specs=backbone.output_specs, min_level=3, max_level=7)
head = segmentation_heads.SegmentationHead(num_classes, level=3)
model = segmentation_model.SegmentationModel(
backbone=backbone,
decoder=decoder,
head=head
)
config = model.get_config()
new_model = segmentation_model.SegmentationModel.from_config(config)
# Validate that the config can be forced to JSON.
_ = new_model.to_json()
# If the serialization was successful, the new config should match the old.
self.assertAllEqual(model.get_config(), new_model.get_config())
def test_segmentation_network_creation(
self, input_size, level):
"""Test for creation of a segmentation network."""
num_classes = 10
inputs = np.random.rand(2, input_size, input_size, 3)
tf.keras.backend.set_image_data_format('channels_last')
backbone = backbones.ResNet(model_id=50)
decoder = fpn.FPN(
input_specs=backbone.output_specs, min_level=2, max_level=7)
head = segmentation_heads.SegmentationHead(num_classes, level=level)
model = segmentation_model.SegmentationModel(
backbone=backbone,
decoder=decoder,
head=head,
mask_scoring_head=None,
)
outputs = model(inputs)
self.assertAllEqual(
[2, input_size // (2**level), input_size // (2**level), num_classes],
outputs['logits'].numpy().shape)
def test_panoptic_deeplab_network_creation(self, input_size, level,
low_level, shared_decoder,
training):
"""Test for creation of a panoptic deeplab network."""
batch_size = 2 if training else 1
num_classes = 10
inputs = np.random.rand(batch_size, input_size, input_size, 3)
image_info = tf.convert_to_tensor([[[input_size, input_size],
[input_size, input_size], [1, 1],
[0, 0]]])
image_info = tf.tile(image_info, [batch_size, 1, 1])
tf.keras.backend.set_image_data_format('channels_last')
backbone = backbones.ResNet(model_id=50)
semantic_decoder = aspp.ASPP(level=level, dilation_rates=[6, 12, 18])
if shared_decoder:
instance_decoder = semantic_decoder
else:
instance_decoder = aspp.ASPP(level=level,
dilation_rates=[6, 12, 18])
semantic_head = panoptic_deeplab_heads.SemanticHead(
num_classes,
level=level,
low_level=low_level,
low_level_num_filters=(64, 32))
instance_head = panoptic_deeplab_heads.InstanceHead(
level=level, low_level=low_level, low_level_num_filters=(64, 32))
post_processor = panoptic_deeplab_merge.PostProcessor(
output_size=[input_size, input_size],
center_score_threshold=0.1,
thing_class_ids=[1, 2, 3, 4],
label_divisor=[256],
stuff_area_limit=4096,
ignore_label=0,
nms_kernel=41,
keep_k_centers=41,
rescale_predictions=True)
model = panoptic_deeplab_model.PanopticDeeplabModel(
backbone=backbone,
semantic_decoder=semantic_decoder,
instance_decoder=instance_decoder,
semantic_head=semantic_head,
instance_head=instance_head,
post_processor=post_processor)
outputs = model(inputs=inputs,
image_info=image_info,
training=training)
if training:
self.assertIn('segmentation_outputs', outputs)
self.assertIn('instance_centers_heatmap', outputs)
self.assertIn('instance_centers_offset', outputs)
self.assertAllEqual([
2, input_size // (2**low_level[-1]), input_size //
(2**low_level[-1]), num_classes
], outputs['segmentation_outputs'].numpy().shape)
self.assertAllEqual([
2, input_size // (2**low_level[-1]), input_size //
(2**low_level[-1]), 1
], outputs['instance_centers_heatmap'].numpy().shape)
self.assertAllEqual([
2, input_size // (2**low_level[-1]), input_size //
(2**low_level[-1]), 2
], outputs['instance_centers_offset'].numpy().shape)
else:
self.assertIn('panoptic_outputs', outputs)
self.assertIn('category_mask', outputs)
self.assertIn('instance_mask', outputs)
self.assertIn('instance_centers', outputs)
self.assertIn('instance_scores', outputs)
self.assertIn('segmentation_outputs', outputs)
