def __init__(self,
scene_json,
datum_names=None,
requested_annotations=None,
requested_autolabels=None,
backward_context=0,
forward_context=0,
generate_depth_from_datum=None,
only_annotated_datums=False):
# Extract a single scene from the scene JSON
scene, calibration_table = BaseSceneDataset._extract_scene_from_scene_json(
scene_json, requested_autolabels=requested_autolabels)
# Return SynchronizedDataset with scenes built from dataset.json
dataset_metadata = DatasetMetadata.from_scene_containers([scene])
super().__init__(dataset_metadata,
scenes=[scene],
calibration_table=calibration_table,
datum_names=datum_names,
requested_annotations=requested_annotations,
requested_autolabels=requested_autolabels,
backward_context=backward_context,
forward_context=forward_context,
generate_depth_from_datum=generate_depth_from_datum,
is_scene_dataset=True,
only_annotated_datums=only_annotated_datums)
def __init__(self,
scene_dataset_json,
split='train',
datum_names=None,
requested_annotations=None,
requested_autolabels=None,
backward_context=0,
forward_context=0,
generate_depth_from_datum=None,
only_annotated_datums=False):
# Extract all scenes from the scene dataset JSON for the appropriate split
scenes, calibration_table = BaseSceneDataset._extract_scenes_from_scene_dataset_json(
scene_dataset_json,
split=split,
requested_autolabels=requested_autolabels)
# Return SynchronizedDataset with scenes built from dataset.json
dataset_metadata = DatasetMetadata.from_scene_containers(scenes)
super().__init__(dataset_metadata,
scenes=scenes,
calibration_table=calibration_table,
datum_names=datum_names,
requested_annotations=requested_annotations,
requested_autolabels=requested_autolabels,
backward_context=backward_context,
forward_context=forward_context,
generate_depth_from_datum=generate_depth_from_datum,
is_scene_dataset=True,
only_annotated_datums=only_annotated_datums)
def __init__(
self,
scene_dataset_json,
split='train',
datum_names=None,
requested_annotations=None,
requested_autolabels=None,
backward_context=0,
forward_context=0,
generate_depth_from_datum=None,
only_annotated_datums=False,
use_virtual_camera_datums=True,
skip_missing_data=False,
accumulation_context=None,
dataset_root=None,
transform_accumulated_box_points=False,
use_diskcache=True,
autolabel_root=None,
):
if not use_diskcache:
logging.warning('Instantiating a dataset with use_diskcache=False may exhaust memory with a large dataset.')
# Extract all scenes from the scene dataset JSON for the appropriate split
scenes = BaseDataset._extract_scenes_from_scene_dataset_json(
scene_dataset_json,
split,
requested_autolabels,
is_datums_synchronized=True,
skip_missing_data=skip_missing_data,
dataset_root=dataset_root,
use_diskcache=use_diskcache,
autolabel_root=autolabel_root,
)
# Return SynchronizedDataset with scenes built from dataset.json
dataset_metadata = DatasetMetadata.from_scene_containers(
scenes,
requested_annotations,
requested_autolabels,
autolabel_root=autolabel_root,
)
super().__init__(
dataset_metadata,
scenes=scenes,
datum_names=datum_names,
requested_annotations=requested_annotations,
requested_autolabels=requested_autolabels,
backward_context=backward_context,
forward_context=forward_context,
generate_depth_from_datum=generate_depth_from_datum,
only_annotated_datums=only_annotated_datums,
use_virtual_camera_datums=use_virtual_camera_datums,
accumulation_context=accumulation_context,
transform_accumulated_box_points=transform_accumulated_box_points,
autolabel_root=autolabel_root,
)
def get_dataset_split(dataset_path):
"""Get a list of splits in the dataset.json.
Parameters
----------
dataset_path: str
Full path to the dataset json holding dataset metadata, ontology, and image and annotation paths.
Returns
-------
dataset_splits: list of str
List of dataset splits (train | val | test | train_overfit).
"""
return DatasetMetadata.get_dataset_splits(dataset_path)