def build(input_reader_config,
model_config,
training,
voxel_generator,
target_assigner=None,
multi_gpu=False) -> DatasetWrapper:
"""Builds a tensor dictionary based on the InputReader config.
Args:
input_reader_config: A input_reader_pb2.InputReader object.
Returns:
A tensor dict based on the input_reader_config.
Raises:
ValueError: On invalid input reader proto.
ValueError: If no input paths are specified.
"""
if not isinstance(input_reader_config, input_reader_pb2.InputReader):
raise ValueError('input_reader_config not of type '
'input_reader_pb2.InputReader.')
dataset = dataset_builder.build(input_reader_config,
model_config,
training,
voxel_generator,
target_assigner,
multi_gpu=multi_gpu) #返回一个数据集
dataset = DatasetWrapper(dataset)
return dataset
def build(input_reader_config,
model_config,
training,
voxel_generator,
target_assigner=None,
net=None,
multi_gpu=False) -> Dataset:
"""Builds a tensor dictionary based on the InputReader config.
Args:
input_reader_config: A input_reader_pb2.InputReader object.
Returns:
A tensor dict based on the input_reader_config.
Raises:
ValueError: On invalid input reader proto.
ValueError: If no input paths are specified.
"""
if not isinstance(input_reader_config, input_reader_pb2.InputReader):
raise ValueError('input_reader_config not of type '
'input_reader_pb2.InputReader.')
dataset = dataset_builder.build(input_reader_config,
model_config,
training,
voxel_generator,
target_assigner,
multi_gpu=multi_gpu)
cum_lc_config = input_reader_config.cum_lc_wrapper
if cum_lc_config.lc_policy != "": # this means that it needs to be wrapped
dataset = CumLCDatasetWrapper(dataset,
net,
cum_lc_config.lc_policy,
cum_lc_config.lc_horizon,
cum_lc_config.init_lidar_num_beams,
cum_lc_config.sparsify_return,
use_cache=not training,
contiguous=not training)
else:
dataset = LidarDatasetWrapper(
dataset, num_workers=input_reader_config.preprocess.num_workers)
return dataset
def build(input_reader_config,
model_config,
training,
voxel_generator,
target_assigner=None,
multi_gpu=False,
generate_anchors_cachae=True, #True for pillar and second
segmentation=False,
bcl_keep_voxels=None,
seg_keep_points=None,
points_per_voxel=None) -> DatasetWrapper:
"""Builds a tensor dictionary based on the InputReader config.
Args:
input_reader_config: A input_reader_pb2.InputReader object.
Returns:
A tensor dict based on the input_reader_config.
Raises:
ValueError: On invalid input reader proto.
ValueError: If no input paths are specified.
"""
if not isinstance(input_reader_config, input_reader_pb2.InputReader):
raise ValueError('input_reader_config not of type '
'input_reader_pb2.InputReader.')
dataset = dataset_builder.build(
input_reader_config,
model_config,
training,
voxel_generator,
target_assigner,
multi_gpu=multi_gpu,
generate_anchors_cachae=generate_anchors_cachae,
segmentation=segmentation,
bcl_keep_voxels=bcl_keep_voxels,
seg_keep_points=seg_keep_points,
points_per_voxel=points_per_voxel) #f use cachae save memory but not suit for BCL
dataset = DatasetWrapper(dataset)
return dataset