def parse_args():
"""
Parse python script parameters (common part).
Returns
-------
ArgumentParser
Resulted args.
"""
parser = argparse.ArgumentParser(
description="Evaluate a model for image matching (PyTorch/HPatches)",
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument("--dataset",
type=str,
default="HPatches",
help="dataset name")
parser.add_argument(
"--work-dir",
type=str,
default=os.path.join("..", "imgclsmob_data"),
help="path to working directory only for dataset root path preset")
args, _ = parser.parse_known_args()
dataset_metainfo = get_dataset_metainfo(dataset_name=args.dataset)
dataset_metainfo.add_dataset_parser_arguments(parser=parser,
work_dir_path=args.work_dir)
add_eval_parser_arguments(parser)
args = parser.parse_args()
return args
def parse_args():
"""
Parse python script parameters (common part).
Returns
-------
ArgumentParser
Resulted args.
"""
parser = argparse.ArgumentParser(
description="Train a model for image classification (PyTorch/FDV)",
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument("--dataset",
type=str,
default="FDV1",
help="dataset name. option are FDV1, FDV2")
parser.add_argument(
"--work-dir",
type=str,
default=os.path.join("..", "facedetver_data"),
help="path to working directory only for dataset root path preset")
args, _ = parser.parse_known_args()
dataset_metainfo = get_dataset_metainfo(dataset_name=args.dataset)
dataset_metainfo.add_dataset_parser_arguments(parser=parser,
work_dir_path=args.work_dir)
add_train_cls_parser_arguments(parser)
args = parser.parse_args()
return args
def parse_args():
"""
Parse python script parameters (common part).
Returns:
-------
ArgumentParser
Resulted args.
"""
parser = argparse.ArgumentParser(
description="Evaluate a model for image classification/segmentation (PyTorch)",
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument(
"--dataset",
type=str,
default="ImageNet1K",
help="dataset name. options are ImageNet1K, CUB200_2011, CIFAR10, CIFAR100, SVHN, VOC2012, ADE20K, Cityscapes, "
"COCO, LibriSpeech, MCV")
parser.add_argument(
"--work-dir",
type=str,
default=os.path.join("..", "imgclsmob_data"),
help="path to working directory only for dataset root path preset")
args, _ = parser.parse_known_args()
dataset_metainfo = get_dataset_metainfo(dataset_name=args.dataset)
dataset_metainfo.add_dataset_parser_arguments(
parser=parser,
work_dir_path=args.work_dir)
add_eval_cls_parser_arguments(parser)
args = parser.parse_args()
return args
def parse_args():
parser = argparse.ArgumentParser(
description="Train a model for image classification (PyTorch)",
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument(
"--dataset",
type=str,
default="ImageNet1K",
help="dataset name. options are ImageNet1K, CUB200_2011, CIFAR10, CIFAR100, SVHN")
parser.add_argument(
"--work-dir",
type=str,
default=os.path.join("..", "imgclsmob_data"),
help="path to working directory only for dataset root path preset")
args, _ = parser.parse_known_args()
dataset_metainfo = get_dataset_metainfo(dataset_name=args.dataset)
dataset_metainfo.add_dataset_parser_arguments(
parser=parser,
work_dir_path=args.work_dir)
add_train_cls_parser_arguments(parser)
args = parser.parse_args()
return args
def prepare_dataset_metainfo(args):
"""
Get dataset metainfo by name of dataset.
Parameters:
----------
args : ArgumentParser
Main script arguments.
Returns:
-------
DatasetMetaInfo
Dataset metainfo.
"""
ds_metainfo = get_dataset_metainfo(dataset_name=args.dataset)
ds_metainfo.update(args=args)
assert (ds_metainfo.ml_type != "imgseg") or (args.batch_size == 1)
assert (ds_metainfo.ml_type != "imgseg") or args.disable_cudnn_autotune
return ds_metainfo
def main():
"""
Main body of script.
"""
args = parse_args()
os.environ["MXNET_CUDNN_AUTOTUNE_DEFAULT"] = "0"
assert (args.batch_size == 1)
_, log_file_exist = initialize_logging(
logging_dir_path=args.save_dir,
logging_file_name=args.logging_file_name,
script_args=args,
log_packages=args.log_packages,
log_pip_packages=args.log_pip_packages)
ds_metainfo = get_dataset_metainfo(dataset_name=args.dataset)
ds_metainfo.update(args=args)
use_cuda, batch_size = prepare_pt_context(
num_gpus=args.num_gpus,
batch_size=args.batch_size)
net = prepare_model(
model_name=args.model,
use_pretrained=args.use_pretrained,
pretrained_model_file_path=args.resume.strip(),
use_cuda=use_cuda,
net_extra_kwargs=ds_metainfo.net_extra_kwargs,
load_ignore_extra=False,
num_classes=args.num_classes,
in_channels=args.in_channels,
remove_module=False)
test_data = get_val_data_source(
ds_metainfo=ds_metainfo,
batch_size=args.batch_size,
num_workers=args.num_workers)
calc_detector_repeatability(
test_data=test_data,
net=net,
use_cuda=use_cuda)
def main():
"""
Main body of script.
"""
args = parse_args()
_, log_file_exist = initialize_logging(
logging_dir_path=args.save_dir,
logging_file_name=args.logging_file_name,
script_args=args,
log_packages=args.log_packages,
log_pip_packages=args.log_pip_packages)
ds_metainfo = get_dataset_metainfo(dataset_name=args.dataset)
ds_metainfo.update(args=args)
use_cuda, batch_size = prepare_pt_context(
num_gpus=args.num_gpus,
batch_size=args.batch_size)
net = prepare_model(
model_name=args.model,
use_pretrained=args.use_pretrained,
pretrained_model_file_path=args.resume.strip(),
use_cuda=use_cuda,
net_extra_kwargs=ds_metainfo.net_extra_kwargs,
load_ignore_extra=ds_metainfo.load_ignore_extra,
num_classes=args.num_classes,
in_channels=args.in_channels,
remove_module=args.remove_module)
real_net = net.module if hasattr(net, "module") else net
input_image_size = real_net.in_size[0] if hasattr(real_net, "in_size") else args.input_size
if args.data_subset == "val":
get_test_data_source_class = get_val_data_source
test_metric = get_composite_metric(
metric_names=ds_metainfo.val_metric_names,
metric_extra_kwargs=ds_metainfo.val_metric_extra_kwargs)
else:
get_test_data_source_class = get_test_data_source
test_metric = get_composite_metric(
metric_names=ds_metainfo.test_metric_names,
metric_extra_kwargs=ds_metainfo.test_metric_extra_kwargs)
test_data = get_test_data_source_class(
ds_metainfo=ds_metainfo,
batch_size=args.batch_size,
num_workers=args.num_workers)
if not args.not_show_progress:
test_data = tqdm(test_data)
assert (args.use_pretrained or args.resume.strip() or args.calc_flops_only)
test(
net=net,
test_data=test_data,
metric=test_metric,
use_cuda=use_cuda,
input_image_size=(input_image_size, input_image_size),
in_channels=args.in_channels,
calc_weight_count=True,
calc_flops=args.calc_flops,
calc_flops_only=args.calc_flops_only,
extended_log=True,
show_bad_samples=args.show_bad_samples)
def main():
"""
Main body of script.
"""
args = parse_args()
if args.disable_cudnn_autotune:
os.environ["MXNET_CUDNN_AUTOTUNE_DEFAULT"] = "0"
_, log_file_exist = initialize_logging(
logging_dir_path=args.save_dir,
logging_file_name=args.logging_file_name,
script_args=args,
log_packages=args.log_packages,
log_pip_packages=args.log_pip_packages)
ds_metainfo = get_dataset_metainfo(dataset_name=args.dataset)
ds_metainfo.update(args=args)
assert (ds_metainfo.ml_type != "imgseg") or (args.batch_size == 1)
assert (ds_metainfo.ml_type != "imgseg") or args.disable_cudnn_autotune
use_cuda, batch_size = prepare_pt_context(
num_gpus=args.num_gpus,
batch_size=args.batch_size)
net = prepare_model(
model_name=args.model,
use_pretrained=args.use_pretrained,
pretrained_model_file_path=args.resume.strip(),
use_cuda=use_cuda,
net_extra_kwargs=ds_metainfo.net_extra_kwargs,
load_ignore_extra=ds_metainfo.load_ignore_extra,
num_classes=args.num_classes,
in_channels=args.in_channels,
remove_module=args.remove_module)
real_net = net.module if hasattr(net, "module") else net
input_image_size = real_net.in_size[0] if hasattr(real_net, "in_size") else args.input_size
if args.data_subset == "val":
get_test_data_source_class = get_val_data_source
test_metric = get_composite_metric(
metric_names=ds_metainfo.val_metric_names,
metric_extra_kwargs=ds_metainfo.val_metric_extra_kwargs)
else:
get_test_data_source_class = get_test_data_source
test_metric = get_composite_metric(
metric_names=ds_metainfo.test_metric_names,
metric_extra_kwargs=ds_metainfo.test_metric_extra_kwargs)
test_data = get_test_data_source_class(
ds_metainfo=ds_metainfo,
batch_size=args.batch_size,
num_workers=args.num_workers)
if args.show_progress:
from tqdm import tqdm
test_data = tqdm(test_data)
assert (args.use_pretrained or args.resume.strip() or args.calc_flops_only)
test(
net=net,
test_data=test_data,
metric=test_metric,
use_cuda=use_cuda,
input_image_size=(input_image_size, input_image_size),
in_channels=args.in_channels,
# calc_weight_count=(not log_file_exist),
calc_weight_count=True,
calc_flops=args.calc_flops,
calc_flops_only=args.calc_flops_only,
extended_log=True)
def main():
"""
Main body of script.
"""
args = parse_args()
args.seed = init_rand(seed=args.seed)
_, log_file_exist = initialize_logging(
logging_dir_path=args.save_dir,
logging_file_name=args.logging_file_name,
script_args=args,
log_packages=args.log_packages,
log_pip_packages=args.log_pip_packages)
use_cuda, batch_size = prepare_pt_context(num_gpus=args.num_gpus,
batch_size=args.batch_size)
net = prepare_model(model_name=args.model,
use_pretrained=args.use_pretrained,
pretrained_model_file_path=args.resume.strip(),
use_cuda=use_cuda)
real_net = net.module if hasattr(net, "module") else net
assert (hasattr(real_net, "num_classes"))
num_classes = real_net.num_classes
ds_metainfo = get_dataset_metainfo(dataset_name=args.dataset)
ds_metainfo.update(args=args)
train_data = get_train_data_source(ds_metainfo=ds_metainfo,
batch_size=batch_size,
num_workers=args.num_workers)
val_data = get_val_data_source(ds_metainfo=ds_metainfo,
batch_size=batch_size,
num_workers=args.num_workers)
optimizer, lr_scheduler, start_epoch = prepare_trainer(
net=net,
optimizer_name=args.optimizer_name,
wd=args.wd,
momentum=args.momentum,
lr_mode=args.lr_mode,
lr=args.lr,
lr_decay_period=args.lr_decay_period,
lr_decay_epoch=args.lr_decay_epoch,
lr_decay=args.lr_decay,
num_epochs=args.num_epochs,
state_file_path=args.resume_state)
if args.save_dir and args.save_interval:
param_names = ds_metainfo.val_metric_capts + ds_metainfo.train_metric_capts + [
"Train.Loss", "LR"
]
lp_saver = TrainLogParamSaver(
checkpoint_file_name_prefix="{}_{}".format(ds_metainfo.short_label,
args.model),
last_checkpoint_file_name_suffix="last",
best_checkpoint_file_name_suffix=None,
last_checkpoint_dir_path=args.save_dir,
best_checkpoint_dir_path=None,
last_checkpoint_file_count=2,
best_checkpoint_file_count=2,
checkpoint_file_save_callback=save_params,
checkpoint_file_exts=(".pth", ".states"),
save_interval=args.save_interval,
num_epochs=args.num_epochs,
param_names=param_names,
acc_ind=ds_metainfo.saver_acc_ind,
# bigger=[True],
# mask=None,
score_log_file_path=os.path.join(args.save_dir, "score.log"),
score_log_attempt_value=args.attempt,
best_map_log_file_path=os.path.join(args.save_dir, "best_map.log"))
else:
lp_saver = None
train_net(batch_size=batch_size,
num_epochs=args.num_epochs,
start_epoch1=args.start_epoch,
train_data=train_data,
val_data=val_data,
net=net,
optimizer=optimizer,
lr_scheduler=lr_scheduler,
lp_saver=lp_saver,
log_interval=args.log_interval,
num_classes=num_classes,
val_metric=get_composite_metric(
ds_metainfo.val_metric_names,
ds_metainfo.val_metric_extra_kwargs),
train_metric=get_composite_metric(
ds_metainfo.train_metric_names,
ds_metainfo.train_metric_extra_kwargs),
use_cuda=use_cuda)
def main():
"""
Main body of script.
"""
args = parse_args()
args.seed = init_rand(seed=args.seed)
# args.num_gpus = 1
# args.num_epochs = 500
# args.dataset = "CIFAR10"
# args.model = "resnet20_cifar10"
args.use_pretrained = False
num_non_res = 3
pretrained_model_file_path = '/exdrive/resnet20-cifar10/unfrozen-training/by-stack/non-res-stack-3-models/cifar10-non-res-stack-2-teacher.pth'
_, log_file_exist = initialize_logging(
logging_dir_path=args.save_dir,
logging_file_name=args.logging_file_name,
script_args=args,
log_packages=args.log_packages,
log_pip_packages=args.log_pip_packages)
use_cuda, batch_size = prepare_pt_context(num_gpus=args.num_gpus,
batch_size=args.batch_size)
# Get model
net = prepare_model(
model_name=args.model,
use_pretrained=args.use_pretrained,
# pretrained_model_file_path=pretrained_model_file_path,
pretrained_model_file_path=args.resume.strip(),
use_cuda=use_cuda,
remove_module=True
) # True if using our own trained model waits; o.w. False
real_net = net.module if hasattr(net, "module") else net
assert (hasattr(real_net, "num_classes"))
num_classes = real_net.num_classes
ds_metainfo = get_dataset_metainfo(dataset_name=args.dataset)
ds_metainfo.update(args=args)
train_data = get_train_data_source(ds_metainfo=ds_metainfo,
batch_size=batch_size,
num_workers=args.num_workers)
val_data = get_val_data_source(ds_metainfo=ds_metainfo,
batch_size=batch_size,
num_workers=args.num_workers)
optimizer, lr_scheduler, start_epoch = prepare_trainer(
net=net,
optimizer_name=args.optimizer_name,
wd=args.wd,
momentum=args.momentum,
lr_mode=args.lr_mode,
lr=args.lr,
lr_decay_period=args.lr_decay_period,
lr_decay_epoch=args.lr_decay_epoch,
lr_decay=args.lr_decay,
num_epochs=args.num_epochs,
state_file_path=args.resume_state)
if args.save_dir and args.save_interval:
param_names = ds_metainfo.val_metric_capts + ds_metainfo.train_metric_capts + [
"Train.Loss", "LR"
]
lp_saver = TrainLogParamSaver(
checkpoint_file_name_prefix="{}_{}".format(ds_metainfo.short_label,
args.model),
last_checkpoint_file_name_suffix="last",
best_checkpoint_file_name_suffix=None,
last_checkpoint_dir_path=args.save_dir,
best_checkpoint_dir_path=None,
last_checkpoint_file_count=2,
best_checkpoint_file_count=2,
checkpoint_file_save_callback=save_params,
checkpoint_file_exts=(".pth", ".states"),
save_interval=args.save_interval,
num_epochs=args.num_epochs,
param_names=param_names,
acc_ind=ds_metainfo.saver_acc_ind,
# bigger=[True],
# mask=None,
score_log_file_path=os.path.join(args.save_dir, "score.log"),
score_log_attempt_value=args.attempt,
best_map_log_file_path=os.path.join(args.save_dir, "best_map.log"))
else:
lp_saver = None
print(net)
# summary(net, (3, 32, 32))
if torch.cuda.is_available():
print('cuda available - sending net to gpu')
print('training using cuda = ', use_cuda)
net.cuda()
print('\n\nTraining nonresnet20 on cifar10 with num_non_res=', num_non_res)
train_net(batch_size=batch_size,
num_epochs=args.num_epochs,
start_epoch1=args.start_epoch,
train_data=train_data,
val_data=val_data,
net=net,
optimizer=optimizer,
lr_scheduler=lr_scheduler,
lp_saver=lp_saver,
log_interval=args.log_interval,
num_classes=num_classes,
val_metric=get_composite_metric(
ds_metainfo.val_metric_names,
ds_metainfo.val_metric_extra_kwargs),
train_metric=get_composite_metric(
ds_metainfo.train_metric_names,
ds_metainfo.train_metric_extra_kwargs),
use_cuda=use_cuda)
