def download_data_sources(self, only_meta=False):
self.logger.info("download_data_sources started")
data_sources = sly.get_data_sources(self.info['graph'])
for proj, datasets in data_sources.items():
pr_name = proj
pr_proto = self.api.simple_request(
'GetProjectByName', api_proto.Project,
api_proto.Project(title=pr_name))
if pr_proto.id == -1:
self.logger.critical('Project not found',
extra={'project_name': pr_name})
raise RuntimeError('Project not found')
datasets_proto_arr = []
if datasets != "*":
for ds_name in datasets:
ds_proto = self.api.simple_request(
'GetDatasetByName', api_proto.Dataset,
api_proto.ProjectDataset(
project=api_proto.Project(id=pr_proto.id),
dataset=api_proto.Dataset(title=ds_name)))
if ds_proto.id == -1:
self.logger.critical('Dataset not found',
extra={
'project_id': pr_proto.id,
'project_title': pr_name,
'dataset_title': ds_name
})
raise RuntimeError('Dataset not found')
datasets_proto_arr.append(
api_proto.Dataset(id=ds_proto.id, title=ds_name))
else:
datasets_proto = self.api.simple_request(
'GetProjectDatasets', api_proto.DatasetArray,
api_proto.Id(id=pr_proto.id))
datasets_proto_arr = datasets_proto.datasets
if only_meta is True:
project_info = self.api.simple_request(
'GetProjectMeta', api_proto.Project,
api_proto.Id(id=pr_proto.id))
pr_writer = sly.ProjectWriterFS(self.dir_data,
project_info.title)
pr_meta = sly.ProjectMeta(sly.json_loads(project_info.meta))
pr_writer.write_meta(pr_meta)
else:
self.data_mgr.download_project(
self.dir_data,
pr_proto,
datasets_proto_arr,
download_images=self.download_images)
def download_step(self):
self.logger.info("DOWNLOAD_DATA")
sly.json_dump(self.info['config'], self.config_path)
pr_info = self.info['project']
project = api_proto.Project(id=pr_info['id'], title=pr_info['title'])
datasets = [
api_proto.Dataset(id=ds['id'], title=ds['title'])
for ds in pr_info['datasets']
]
self.data_mgr.download_project(self.dir_data, project, datasets)
if self.info.get('nn_model', None) is not None:
nn_id = self.info['nn_model']['id']
nn_hash = self.info['nn_model']['hash']
self.logger.info('DOWNLOAD_NN',
extra={
'nn_id': nn_id,
'nn_hash': nn_hash
})
self.data_mgr.download_nn(nn_id, nn_hash, self.dir_model)
else:
self.logger.info('Initializing task without source NN.')
self.report_step_done(TaskStep.DOWNLOAD)
def _create_dataset(self, project_id, dataset_name):
remote_name = dataset_name
for _ in range(3):
dataset = self.api.simple_request('CreateDataset',
api_proto.Id,
api_proto.ProjectDataset(project=api_proto.Project(id=project_id),
dataset=api_proto.Dataset(title=remote_name)))
if dataset.id != 0: # created
return dataset.id, remote_name
remote_name = '{}_{}'.format(dataset_name, sly.generate_random_string(5))
raise RuntimeError('Unable to create dataset with random suffix.')
def download_step(self):
if self.info.get('nn_model', None) is None:
self.logger.critical('TASK_NN_EMPTY')
raise ValueError('TASK_NN_EMPTY')
nn_id = self.info['nn_model']['id']
nn_hash = self.info['nn_model']['hash']
self.logger.info('DOWNLOAD_NN', extra={'nn_id': nn_id, 'nn_hash': nn_hash})
self.data_mgr.download_nn(nn_id, nn_hash, self.dir_model)
self.logger.info("DOWNLOAD_DATA")
sly.json_dump(self.info['config'], self.config_path)
pr_info = self.info['project']
project = api_proto.Project(id=pr_info['id'], title=pr_info['title'])
datasets = [api_proto.Dataset(id=ds['id'], title=ds['title']) for ds in pr_info['datasets']]
self.data_mgr.download_project(self.dir_data, project, datasets)
self.report_step_done(TaskStep.DOWNLOAD)