def test_mixup_transform_single_label_missing_num_classes(self):
alpha = 2.0
mixup_transform = MixupTransform(alpha, None)
sample = {
"input": torch.rand(4, 3, 224, 224, dtype=torch.float32),
"target": torch.as_tensor([0, 1, 2, 2], dtype=torch.int32),
}
with self.assertRaises(Exception):
mixup_transform(sample)
def test_mixup_transform_single_label(self):
alpha = 2.0
num_classes = 3
mixup_transform = MixupTransform(alpha, num_classes)
sample = {
"input": torch.rand(4, 3, 224, 224, dtype=torch.float32),
"target": torch.as_tensor([0, 1, 2, 2], dtype=torch.int32),
}
sample_mixup = mixup_transform(sample)
self.assertTrue(sample["input"].shape == sample_mixup["input"].shape)
self.assertTrue(sample_mixup["target"].shape[0] == 4)
self.assertTrue(sample_mixup["target"].shape[1] == 3)
def test_mixup_transform_single_label_multi_modal_batch(self):
mixup_alpha = 2.0
num_classes = 3
mixup_transform = MixupTransform(mixup_alpha, num_classes)
sample = {
"input": {
"video": torch.rand(4, 3, 4, 224, 224, dtype=torch.float32),
"audio": torch.rand(4, 1, 40, 100, dtype=torch.float32),
},
"target": torch.as_tensor([0, 1, 2, 2], dtype=torch.int32),
}
mixup_transform(sample)
def test_mixup_transform_multi_label(self):
alpha = 2.0
mixup_transform = MixupTransform(alpha, None)
sample = {
"input": torch.rand(4, 3, 224, 224, dtype=torch.float32),
"target": torch.as_tensor(
[[1, 0, 0, 0], [0, 1, 0, 1], [0, 0, 1, 1], [0, 1, 1, 1]],
dtype=torch.int32,
),
}
sample_mixup = mixup_transform(sample)
self.assertTrue(sample["input"].shape == sample_mixup["input"].shape)
self.assertTrue(sample["target"].shape == sample_mixup["target"].shape)
def test_mixup_transform_single_label_image_batch(self):
mixup_alpha = 2.0
num_classes = 3
for mode in ["batch", "pair", "elem"]:
mixup_transform = MixupTransform(mixup_alpha, num_classes, mode=mode)
sample = {
"input": torch.rand(4, 3, 224, 224, dtype=torch.float32),
"target": torch.as_tensor([0, 1, 2, 2], dtype=torch.int32),
}
sample_mixup = mixup_transform(sample)
self.assertTrue(sample["input"].shape == sample_mixup["input"].shape)
self.assertTrue(sample_mixup["target"].shape[0] == 4)
self.assertTrue(sample_mixup["target"].shape[1] == 3)
def test_mixup_transform_multi_label_multi_modal_batch(self):
mixup_alpha = 2.0
mixup_transform = MixupTransform(mixup_alpha, None)
sample = {
"input": {
"video": torch.rand(4, 3, 4, 224, 224, dtype=torch.float32),
"audio": torch.rand(4, 1, 40, 100, dtype=torch.float32),
},
"target": torch.as_tensor(
[[1, 0, 0, 0], [0, 1, 0, 1], [0, 0, 1, 1], [0, 1, 1, 1]],
dtype=torch.int32,
),
}
mixup_transform(sample)
def test_cutmix_transform_single_label_image_batch(self):
mixup_alpha = 0
cutmix_alpha = 0.2
num_classes = 3
for mode in ["batch", "pair", "elem"]:
for minmax in [None, (0.3, 0.7)]:
cutmix_transform = MixupTransform(
mixup_alpha,
num_classes,
cutmix_alpha=cutmix_alpha,
mode=mode,
cutmix_minmax=minmax,
)
sample = {
"input": torch.rand(4, 3, 224, 224, dtype=torch.float32),
"target": torch.as_tensor([0, 1, 2, 2], dtype=torch.int32),
}
sample_cutmix = cutmix_transform(sample)
self.assertTrue(sample["input"].shape == sample_cutmix["input"].shape)
self.assertTrue(sample_cutmix["target"].shape[0] == 4)
self.assertTrue(sample_cutmix["target"].shape[1] == 3)
def test_mixup_cutmix_transform_single_label_image_batch(self):
mixup_alpha = 0.3
cutmix_alpha = 0.2
num_classes = 3
for mode in ["batch", "pair", "elem"]:
cutmix_transform = MixupTransform(
mixup_alpha,
num_classes,
cutmix_alpha=cutmix_alpha,
switch_prob=0.5,
mode=mode,
)
for _i in range(4):
sample = {
"input": torch.rand(4, 3, 224, 224, dtype=torch.float32),
"target": torch.as_tensor([0, 1, 2, 2], dtype=torch.int32),
}
sample_cutmix = cutmix_transform(sample)
self.assertTrue(sample["input"].shape == sample_cutmix["input"].shape)
self.assertTrue(sample_cutmix["target"].shape[0] == 4)
self.assertTrue(sample_cutmix["target"].shape[1] == 3)
def from_config(cls, config: Dict[str, Any]) -> "ClassificationTask":
"""Instantiates a ClassificationTask from a configuration.
Args:
config: A configuration for a ClassificationTask.
See :func:`__init__` for parameters expected in the config.
Returns:
A ClassificationTask instance.
"""
test_only = config.get("test_only", False)
if not test_only:
# TODO Make distinction between epochs and phases in optimizer clear
train_phases_per_epoch = config["dataset"]["train"].get(
"phases_per_epoch", 1)
optimizer_config = config["optimizer"]
optimizer_config["num_epochs"] = (config["num_epochs"] *
train_phases_per_epoch)
optimizer = build_optimizer(optimizer_config)
param_schedulers = build_optimizer_schedulers(optimizer_config)
datasets = {}
phase_types = ["train", "test"]
for phase_type in phase_types:
if phase_type in config["dataset"]:
datasets[phase_type] = build_dataset(
config["dataset"][phase_type])
loss = build_loss(config["loss"])
amp_args = config.get("amp_args")
meters = build_meters(config.get("meters", {}))
model = build_model(config["model"])
mixup_transform = None
if config.get("mixup") is not None:
assert "alpha" in config[
"mixup"], "key alpha is missing in mixup dict"
mixup_transform = MixupTransform(
config["mixup"]["alpha"], config["mixup"].get("num_classes"))
# hooks config is optional
hooks_config = config.get("hooks")
hooks = []
if hooks_config is not None:
hooks = build_hooks(hooks_config)
distributed_config = config.get("distributed", {})
distributed_options = {
"broadcast_buffers_mode":
BroadcastBuffersMode[distributed_config.get(
"broadcast_buffers", "before_eval").upper()],
"batch_norm_sync_mode":
BatchNormSyncMode[distributed_config.get("batch_norm_sync_mode",
"disabled").upper()],
"batch_norm_sync_group_size":
distributed_config.get("batch_norm_sync_group_size", 0),
"find_unused_parameters":
distributed_config.get("find_unused_parameters", True),
}
task = (
cls().set_num_epochs(config["num_epochs"]).set_test_phase_period(
config.get(
"test_phase_period",
1)).set_loss(loss).set_test_only(test_only).set_model(
model).set_meters(meters).set_amp_args(amp_args).
set_mixup_transform(mixup_transform).set_distributed_options(
**distributed_options).set_hooks(hooks).set_bn_weight_decay(
config.get("bn_weight_decay", False)))
if not test_only:
task.set_optimizer(optimizer)
task.set_optimizer_schedulers(param_schedulers)
use_gpu = config.get("use_gpu")
if use_gpu is not None:
task.set_use_gpu(use_gpu)
for phase_type in datasets:
task.set_dataset(datasets[phase_type], phase_type)
# NOTE: this is a private member and only meant to be used for
# logging/debugging purposes. See __repr__ implementation
task._config = config
return task
def from_config(cls, config: Dict[str, Any]) -> "ClassificationTask":
"""Instantiates a ClassificationTask from a configuration.
Args:
config: A configuration for a ClassificationTask.
See :func:`__init__` for parameters expected in the config.
Returns:
A ClassificationTask instance.
"""
optimizer_config = config["optimizer"]
# TODO Make distinction between epochs and phases in optimizer clear
train_phases_per_epoch = config["dataset"]["train"].get(
"phases_per_epoch", 1)
optimizer_config[
"num_epochs"] = config["num_epochs"] * train_phases_per_epoch
datasets = {}
phase_types = ["train", "test"]
for phase_type in phase_types:
datasets[phase_type] = build_dataset(config["dataset"][phase_type])
loss = build_loss(config["loss"])
test_only = config.get("test_only", False)
amp_args = config.get("amp_args")
meters = build_meters(config.get("meters", {}))
model = build_model(config["model"])
mixup_transform = None
if config.get("mixup") is not None:
assert "alpha" in config[
"mixup"], "key alpha is missing in mixup dict"
mixup_transform = MixupTransform(
config["mixup"]["alpha"], config["mixup"].get("num_classes"))
# hooks config is optional
hooks_config = config.get("hooks")
hooks = []
if hooks_config is not None:
hooks = build_hooks(hooks_config)
optimizer = build_optimizer(optimizer_config)
task = (cls().set_num_epochs(
config["num_epochs"]).set_test_phase_period(
config.get(
"test_phase_period",
1)).set_loss(loss).set_test_only(test_only).set_model(
model).set_optimizer(optimizer).set_meters(
meters).set_amp_args(amp_args).
set_mixup_transform(mixup_transform).set_distributed_options(
broadcast_buffers_mode=BroadcastBuffersMode[config.get(
"broadcast_buffers", "disabled").upper()],
batch_norm_sync_mode=BatchNormSyncMode[config.get(
"batch_norm_sync_mode", "disabled").upper()],
find_unused_parameters=config.get("find_unused_parameters",
True),
).set_hooks(hooks))
use_gpu = config.get("use_gpu")
if use_gpu is not None:
task.set_use_gpu(use_gpu)
for phase_type in phase_types:
task.set_dataset(datasets[phase_type], phase_type)
return task
