def validate_config(self):
ids = [s.id for s in self.train_scenes]
if len(set(ids)) != len(ids):
raise ConfigError('All training scene ids must be unique.')
ids = [s.id for s in self.validation_scenes + self.test_scenes]
if len(set(ids)) != len(ids):
raise ConfigError(
'All validation and test scene ids must be unique.')
def setup_data(self):
"""Set the the DataSet and DataLoaders for train, validation, and test sets."""
cfg = self.cfg
batch_sz = self.cfg.solver.batch_sz
num_workers = self.cfg.data.num_workers
train_ds, valid_ds, test_ds = self.get_datasets()
if len(train_ds) < batch_sz:
raise ConfigError(
'Training dataset has fewer elements than batch size.')
if len(valid_ds) < batch_sz:
raise ConfigError(
'Validation dataset has fewer elements than batch size.')
if len(test_ds) < batch_sz:
raise ConfigError(
'Test dataset has fewer elements than batch size.')
if cfg.overfit_mode:
train_ds = Subset(train_ds, range(batch_sz))
valid_ds = train_ds
test_ds = train_ds
elif cfg.test_mode:
train_ds = Subset(train_ds, range(batch_sz))
valid_ds = Subset(valid_ds, range(batch_sz))
test_ds = Subset(test_ds, range(batch_sz))
if cfg.data.train_sz is not None:
train_inds = list(range(len(train_ds)))
random.shuffle(train_inds)
train_inds = train_inds[0:cfg.data.train_sz]
train_ds = Subset(train_ds, train_inds)
collate_fn = self.get_collate_fn()
train_dl = DataLoader(train_ds,
shuffle=True,
batch_size=batch_sz,
num_workers=num_workers,
pin_memory=True,
collate_fn=collate_fn)
valid_dl = DataLoader(valid_ds,
shuffle=True,
batch_size=batch_sz,
num_workers=num_workers,
pin_memory=True,
collate_fn=collate_fn)
test_dl = DataLoader(test_ds,
shuffle=True,
batch_size=batch_sz,
num_workers=num_workers,
pin_memory=True,
collate_fn=collate_fn)
self.train_ds, self.valid_ds, self.test_ds = (train_ds, valid_ds,
test_ds)
self.train_dl, self.valid_dl, self.test_dl = (train_dl, valid_dl,
test_dl)
def validate_config(self):
if len(self.class_names) != len(self.class_colors):
raise ConfigError('len(class_names) must equal len(class_colors')
self.validate_nonneg('img_sz')
self.validate_nonneg('num_workers')
self.validate_augmentors()
self.validate_data_format()
def validate_config(self):
ids = [s.id for s in self.get_all_scenes()]
if len(set(ids)) != len(ids):
raise ConfigError('All scene ids must be unique.')
def validate_config(self):
if self.sample_prob > 1 or self.sample_prob <= 0:
raise ConfigError('sample_prob must be <= 1 and > 0')
def validate_config(self):
if self.run_tensorboard and not self.log_tensorboard:
raise ConfigError(
'Cannot run_tensorboard if log_tensorboard is False')
def validate_config(self):
if self.null_class is not None and self.null_class not in self.names:
raise ConfigError(
'The null_class: {} must be in list of class names.'.format(
self.null_class))
def validate_config(self):
if self.train_chip_sz != self.predict_chip_sz:
raise ConfigError(
'train_chip_sz must be equal to predict_chip_sz for chip '
'classification.')
def validate_config(self):
if self.vector_source.has_null_class_bufs():
raise ConfigError(
'Setting buffer to None for a class in the vector_source is '
'not allowed for RasterizedSourceConfig.')
def validate_config(self):
if self.vector_source.has_null_class_bufs():
raise ConfigError(
'Setting buffer to None for a class in the vector_source is '
'not allowed for ChipClassificationLabelSourceConfig.')