def to_proto(self):
d = {
'sync_interval': self.train_options.sync_interval,
'do_monitoring': self.train_options.do_monitoring,
'replace_model': self.train_options.replace_model,
'model_main_py': self.script_locations.model_main_uri,
'export_py': self.script_locations.export_uri,
'training_data_uri': self.training_data_uri,
'training_output_uri': self.training_output_uri,
'model_uri': self.model_uri,
'debug': self.debug,
'fine_tune_checkpoint_name': self.fine_tune_checkpoint_name,
'tfod_config': self.tfod_config
}
conf = json_format.ParseDict(
d, BackendConfigMsg.TFObjectDetectionConfig())
msg = BackendConfigMsg(
backend_type=rv.TF_OBJECT_DETECTION,
tf_object_detection_config=conf)
if self.pretrained_model_uri:
msg.MergeFrom(
BackendConfigMsg(
pretrained_model_uri=self.pretrained_model_uri))
return msg
def to_proto(self):
d = {
'train_py': self.script_locations.train_py,
'eval_py': self.script_locations.eval_py,
'export_py': self.script_locations.export_py,
'train_restart_dir': self.train_options.train_restart_dir,
'sync_interval': self.train_options.sync_interval,
'do_monitoring': self.train_options.do_monitoring,
'do_eval': self.train_options.do_eval,
'replace_model': self.train_options.replace_model,
'debug': self.debug,
'training_data_uri': self.training_data_uri,
'training_output_uri': self.training_output_uri,
'model_uri': self.model_uri,
'fine_tune_checkpoint_name': self.fine_tune_checkpoint_name,
'tfdl_config': self.tfdl_config
}
conf = json_format.ParseDict(d, BackendConfigMsg.TFDeeplabConfig())
msg = BackendConfigMsg(backend_type=rv.TF_DEEPLAB,
tf_deeplab_config=conf)
if self.pretrained_model_uri:
msg.MergeFrom(
BackendConfigMsg(
pretrained_model_uri=self.pretrained_model_uri))
return msg
def test_build_task_from_proto(self):
config = {
'backend_type': rv.TF_OBJECT_DETECTION,
'tf_object_detection_config': {
'tfod_config': {
'model': {
'dummy': {
'numClasses': 5,
'imageResizer': {
'keepAspectRatioResizer': {
'minDimension': 300,
'maxDimension': 300
}
}
}
},
'trainConfig': {
'batchSize': 1,
'numSteps': 200
}
}
}
}
msg = json_format.ParseDict(config, BackendConfigMsg())
b = rv.BackendConfig.from_proto(msg)
self.assertEqual(b.tfod_config['trainConfig']['batchSize'], 1)
self.assertEqual(b.tfod_config['trainConfig']['numSteps'], 200)
def test_build_task_from_proto(self):
config = {
'backend_type': rv.KERAS_CLASSIFICATION,
'keras_classification_config': {
'kc_config': {
'model': {
'input_size': 300,
'type': 'RESNET50',
'load_weights_by_name': False,
'model_path': ''
},
'trainer': {
'optimizer': {
'type': 'ADAM',
'init_lr': 0.0001
},
'options': {
'training_data_dir': '',
'validation_data_dir': '',
'nb_epochs': 1,
'batch_size': 1,
'input_size': 300,
'output_dir': '',
'class_names': ['TEMPLATE'],
'short_epoch': True
}
}
}
}
}
msg = json_format.ParseDict(config, BackendConfigMsg())
b = rv.BackendConfig.from_proto(msg)
self.assertEqual(b.kc_config['model']['type'], 'RESNET50')
def to_proto(self):
result = self.mock.to_proto()
if result is None:
return BackendConfigMsg(backend_type=self.backend_type,
custom_config={})
else:
return result
def to_proto(self):
custom_config = struct_pb2.Struct()
for k, v in self.config.items():
custom_config[k] = v
msg = BackendConfigMsg(
backend_type=self.backend_type,
custom_config=custom_config)
return msg
def to_proto(self):
struct = struct_pb2.Struct()
struct['sync_interval'] = self.train_options.sync_interval
struct['do_monitoring'] = self.train_options.do_monitoring
struct['replace_model'] = self.train_options.replace_model
struct['model_uri'] = self.model_uri
struct['debug'] = self.debug
struct['training_data_uri'] = self.training_data_uri
struct['training_output_uri'] = self.training_output_uri
msg = BackendConfigMsg(backend_type=self.backend_type,
custom_config=struct)
if self.pretrained_model_uri:
msg.MergeFrom(
BackendConfigMsg(
pretrained_model_uri=self.pretrained_model_uri))
return msg
def to_proto(self):
struct = struct_pb2.Struct()
struct['scenes'] = list(
map(lambda x: json_format.MessageToDict(x.to_proto()),
self.scenes))
struct['batch_size'] = self.batch_size
struct['epochs'] = self.epochs
struct['epoch_size'] = self.epoch_size
struct['epoch_save_rate'] = self.epoch_save_rate
struct['training_output_uri'] = self.training_output_uri
struct['model_uri'] = self.model_uri
msg = BackendConfigMsg(backend_type=self.backend_type,
custom_config=struct)
if self.pretrained_model_uri:
msg.MergeFrom(
BackendConfigMsg(
pretrained_model_uri=self.pretrained_model_uri))
return msg
def to_proto(self):
config = {}
for k, v in self.backend_opts.__dict__.items():
config[k] = v
for k, v in self.train_opts.__dict__.items():
config[k] = v
custom_config = struct_pb2.Struct()
custom_config['json'] = json.dumps(config)
msg = BackendConfigMsg(backend_type=self.backend_type,
custom_config=custom_config)
return msg
def to_proto(self):
d = {
'sync_interval': self.train_options.sync_interval,
'do_monitoring': self.train_options.do_monitoring,
'replace_model': self.train_options.replace_model,
'training_data_uri': self.training_data_uri,
'training_output_uri': self.training_output_uri,
'model_uri': self.model_uri,
'debug': self.debug,
'kc_config': self.kc_config
}
conf = json_format.ParseDict(
d, BackendConfigMsg.KerasClassificationConfig())
msg = BackendConfigMsg(backend_type=rv.KERAS_CLASSIFICATION,
keras_classification_config=conf)
if self.pretrained_model_uri:
msg.MergeFrom(
BackendConfigMsg(
pretrained_model_uri=self.pretrained_model_uri))
return msg
def test_build_backend_from_proto(self):
config = {
'backend_type': rv.TF_DEEPLAB,
'tf_deeplab_config': {
'tfdl_config': {
'decoderOutputStride': 2,
'outputStride': 17
}
}
}
msg = json_format.ParseDict(config, BackendConfigMsg())
b = rv.BackendConfig.from_proto(msg)
self.assertEqual(b.tfdl_config['decoderOutputStride'], 2)
self.assertEqual(b.tfdl_config['outputStride'], 17)
def to_proto(self):
msg = BackendConfigMsg(backend_type=self.backend_type,
custom_config={})
return msg
