def build_model(model_name: Text, inputs: tf.Tensor, **kwargs):
"""Build model for a given model name.
Args:
model_name: the name of the model.
inputs: an image tensor or a numpy array.
**kwargs: extra parameters for model builder.
Returns:
(cls_outputs, box_outputs): the outputs for class and box predictions.
Each is a dictionary with key as feature level and value as predictions.
"""
model_arch = det_model_fn.get_model_arch(model_name)
mixed_precision = kwargs.get('mixed_precision', None)
precision = utils.get_precision(kwargs.get('strategy', None),
mixed_precision)
cls_outputs, box_outputs = utils.build_model_with_precision(
precision, model_arch, inputs, False, model_name, **kwargs)
if mixed_precision:
# Post-processing has multiple places with hard-coded float32.
# TODO(tanmingxing): Remove them once post-process can adpat to dtypes.
cls_outputs = {
k: tf.cast(v, tf.float32)
for k, v in cls_outputs.items()
}
box_outputs = {
k: tf.cast(v, tf.float32)
for k, v in box_outputs.items()
}
return cls_outputs, box_outputs
def build_model(model_name: Text, inputs: tf.Tensor, **kwargs):
"""Build model for a given model name.
Args:
model_name: the name of the model.
inputs: an image tensor or a numpy array.
**kwargs: extra parameters for model builder.
Returns:
(cls_outputs, box_outputs): the outputs for class and box predictions.
Each is a dictionary with key as feature level and value as predictions.
"""
mixed_precision = kwargs.get('mixed_precision', None)
precision = utils.get_precision(kwargs.get('strategy', None), mixed_precision)
if kwargs.get('use_keras_model', None):
def model_arch(feats, model_name=None, **kwargs):
"""Construct a model arch for keras models."""
config = hparams_config.get_efficientdet_config(model_name)
config.override(kwargs)
model = efficientdet_keras.EfficientDetNet(config=config)
#l=model.layers[0] # efficientnet part
#print(l.name)
#layer_names=[]
#feats_out=l.predict(feats,steps=1) #predict
#for ml in l.layers:
#print(ml.name)
#layer_names.append(ml.name)
#save_feat_fig(feats_out)
#exit()
cls_out_list, box_out_list = model(feats, training=False)
# convert the list of model outputs to a dictionary with key=level.
assert len(cls_out_list) == config.max_level - config.min_level + 1
assert len(box_out_list) == config.max_level - config.min_level + 1
cls_outputs, box_outputs = {}, {}
for i in range(config.min_level, config.max_level + 1):
cls_outputs[i] = cls_out_list[i - config.min_level]
box_outputs[i] = box_out_list[i - config.min_level]
return cls_outputs, box_outputs
else:
model_arch = det_model_fn.get_model_arch(model_name)
cls_outputs, box_outputs = utils.build_model_with_precision(
precision, model_arch, inputs, False, model_name, **kwargs)
if mixed_precision:
# Post-processing has multiple places with hard-coded float32.
# TODO(tanmingxing): Remove them once post-process can adpat to dtypes.
cls_outputs = {k: tf.cast(v, tf.float32) for k, v in cls_outputs.items()}
box_outputs = {k: tf.cast(v, tf.float32) for k, v in box_outputs.items()}
return cls_outputs, box_outputs
def build_model(model_name: Text, inputs: tf.Tensor, **kwargs):
"""Build model for a given model name.
Args:
model_name: the name of the model.
inputs: an image tensor or a numpy array.
**kwargs: extra parameters for model builder.
Returns:
(class_outputs, box_outputs): the outputs for class and box predictions.
Each is a dictionary with key as feature level and value as predictions.
"""
model_arch = det_model_fn.get_model_arch(model_name)
class_outputs, box_outputs = model_arch(inputs, model_name, **kwargs)
return class_outputs, box_outputs
def build_model(self,
inputs: tf.Tensor,
is_training: bool = False) -> List[tf.Tensor]:
"""Build model with inputs and labels and print out model stats."""
tf.logging.info('start building model')
model_arch = det_model_fn.get_model_arch(self.model_name)
cls_outputs, box_outputs = model_arch(inputs,
model_name=self.model_name,
is_training_bn=is_training,
use_bfloat16=False,
**self.model_overrides)
print('backbone+fpn+box params/flops = {:.6f}M, {:.9f}B'.format(
*utils.num_params_flops()))
all_outputs = list(cls_outputs.values()) + list(box_outputs.values())
return all_outputs
def build_model(self,
inputs: tf.Tensor,
is_training: bool = False) -> List[tf.Tensor]:
"""Build model with inputs and labels and print out model stats."""
logging.info('start building model')
model_arch = det_model_fn.get_model_arch(self.model_name)
cls_outputs, box_outputs = model_arch(inputs,
model_name=self.model_name,
is_training_bn=is_training,
use_bfloat16=False,
**self.model_overrides)
print('backbone+fpn+box params/flops = {:.6f}M, {:.9f}B'.format(
*utils.num_params_flops()))
# Write to tfevent for tensorboard.
train_writer = tf.summary.FileWriter(self.logdir)
train_writer.add_graph(tf.get_default_graph())
train_writer.flush()
all_outputs = list(cls_outputs.values()) + list(box_outputs.values())
return all_outputs
def build_model(model_name: Text, inputs: tf.Tensor, **kwargs):
model_arch = det_model_fn.get_model_arch(model_name)
class_outputs, box_outputs = model_arch(inputs, model_name, **kwargs)
return class_outputs, box_outputs
