def test_class_balance_dataset(self):
# Test build
dataset = build_dataset(
dict(
type='ClassBalancedDataset',
dataset=self.dataset_cfg,
oversample_thr=1.,
))
assert isinstance(dataset, ClassBalancedDataset)
assert dataset.dataset.test_mode == self.dataset_cfg['test_mode']
# Test default_args
dataset = build_dataset(
dict(
type='ClassBalancedDataset',
dataset=self.dataset_cfg,
oversample_thr=1.,
), {'test_mode': True})
assert dataset.dataset.test_mode == self.dataset_cfg['test_mode']
cp_cfg = deepcopy(self.dataset_cfg)
cp_cfg.pop('test_mode')
dataset = build_dataset(
dict(
type='ClassBalancedDataset',
dataset=cp_cfg,
oversample_thr=1.,
), {'test_mode': True})
assert dataset.dataset.test_mode
def test_normal_dataset(self):
# Test build
dataset = build_dataset(self.dataset_cfg)
assert isinstance(dataset, ImageNet)
assert dataset.test_mode == self.dataset_cfg['test_mode']
# Test default_args
dataset = build_dataset(self.dataset_cfg, {'test_mode': True})
assert dataset.test_mode == self.dataset_cfg['test_mode']
cp_cfg = deepcopy(self.dataset_cfg)
cp_cfg.pop('test_mode')
dataset = build_dataset(cp_cfg, {'test_mode': True})
assert dataset.test_mode
def test_concat_dataset(self):
# Test build
dataset = build_dataset([self.dataset_cfg, self.dataset_cfg])
assert isinstance(dataset, ConcatDataset)
assert dataset.datasets[0].test_mode == self.dataset_cfg['test_mode']
# Test default_args
dataset = build_dataset([self.dataset_cfg, self.dataset_cfg],
{'test_mode': True})
assert dataset.datasets[0].test_mode == self.dataset_cfg['test_mode']
cp_cfg = deepcopy(self.dataset_cfg)
cp_cfg.pop('test_mode')
dataset = build_dataset([cp_cfg, cp_cfg], {'test_mode': True})
assert dataset.datasets[0].test_mode
def main():
args = parse_args()
outputs = mmcv.load(args.pkl_results)
assert 'class_scores' in outputs, \
'No "class_scores" in result file, please set "--out-items" in test.py'
cfg = Config.fromfile(args.config)
assert args.metrics, (
'Please specify at least one metric the argument "--metrics".')
if args.cfg_options is not None:
cfg.merge_from_dict(args.cfg_options)
# import modules from string list.
if cfg.get('custom_imports', None):
from mmcv.utils import import_modules_from_strings
import_modules_from_strings(**cfg['custom_imports'])
cfg.data.test.test_mode = True
dataset = build_dataset(cfg.data.test)
pred_score = outputs['class_scores']
kwargs = {} if args.eval_options is None else args.eval_options
eval_kwargs = cfg.get('evaluation', {}).copy()
# hard-code way to remove EvalHook args
for key in [
'interval', 'tmpdir', 'start', 'gpu_collect', 'save_best', 'rule'
]:
eval_kwargs.pop(key, None)
eval_kwargs.update(dict(metric=args.metrics, **kwargs))
print(dataset.evaluate(pred_score, **eval_kwargs))
def test_sample_label(self):
train_dataset = build_dataset(self.cfg.data.train)
for i in tqdm(range(len(train_dataset))):
sample = train_dataset[i]
labels = sample["gt_labels"].numpy().tolist()
for label in labels:
assert 0 <= label <= 66, F"sample index: {i}, labels is: {labels}"
def test_show_train_sample(self):
train_dataset = build_dataset(self.cfg.data.train)
random_index = random.choice(range(len(train_dataset)))
print(F"train random index: {random_index}")
random_train_data = train_dataset[random_index]
print(F"image shape: {random_train_data['img'].data.shape}")
self.show_train_image(random_train_data, "train_sample")
def test_kfold_dataset(self):
# Test build
dataset = build_dataset(
dict(
type='KFoldDataset',
dataset=self.dataset_cfg,
fold=0,
num_splits=5,
test_mode=False,
))
assert isinstance(dataset, KFoldDataset)
assert not dataset.test_mode
assert dataset.dataset.test_mode == self.dataset_cfg['test_mode']
# Test default_args
dataset = build_dataset(dict(
type='KFoldDataset',
dataset=self.dataset_cfg,
fold=0,
num_splits=5,
test_mode=False,
),
default_args={
'test_mode': True,
'classes': [1, 2, 3]
})
assert not dataset.test_mode
assert dataset.dataset.test_mode == self.dataset_cfg['test_mode']
assert dataset.dataset.CLASSES == [1, 2, 3]
cp_cfg = deepcopy(self.dataset_cfg)
cp_cfg.pop('test_mode')
dataset = build_dataset(dict(
type='KFoldDataset',
dataset=self.dataset_cfg,
fold=0,
num_splits=5,
),
default_args={
'test_mode': True,
'classes': [1, 2, 3]
})
# The test_mode in default_args will be passed to KFoldDataset
assert dataset.test_mode
assert not dataset.dataset.test_mode
# Other default_args will be passed to child dataset.
assert dataset.dataset.CLASSES == [1, 2, 3]
def main():
args = parse_args()
if args.out is not None and not args.out.endswith(('.pkl', '.pickle')):
raise ValueError('The output file must be a pkl file.')
cfg = mmcv.Config.fromfile(args.config)
if args.cfg_options is not None:
cfg.merge_from_dict(args.cfg_options)
# build dataset and dataloader
dataset = build_dataset(cfg.data.test)
data_loader = build_dataloader(dataset,
samples_per_gpu=cfg.data.samples_per_gpu,
workers_per_gpu=cfg.data.workers_per_gpu,
shuffle=False,
round_up=False)
# build onnxruntime model and run inference.
if args.backend == 'onnxruntime':
model = ONNXRuntimeClassifier(args.model,
class_names=dataset.CLASSES,
device_id=0)
elif args.backend == 'tensorrt':
model = TensorRTClassifier(args.model,
class_names=dataset.CLASSES,
device_id=0)
else:
print('Unknown backend: {}.'.format(args.model))
exit(1)
model = MMDataParallel(model, device_ids=[0])
model.CLASSES = dataset.CLASSES
outputs = single_gpu_test(model, data_loader, args.show, args.show_dir)
if args.metrics:
results = dataset.evaluate(outputs, args.metrics, args.metric_options)
for k, v in results.items():
print(f'\n{k} : {v:.2f}')
else:
warnings.warn('Evaluation metrics are not specified.')
scores = np.vstack(outputs)
pred_score = np.max(scores, axis=1)
pred_label = np.argmax(scores, axis=1)
pred_class = [dataset.CLASSES[lb] for lb in pred_label]
results = {
'pred_score': pred_score,
'pred_label': pred_label,
'pred_class': pred_class
}
if not args.out:
print('\nthe predicted result for the first element is '
f'pred_score = {pred_score[0]:.2f}, '
f'pred_label = {pred_label[0]} '
f'and pred_class = {pred_class[0]}. '
'Specify --out to save all results to files.')
if args.out:
print(f'\nwriting results to {args.out}')
mmcv.dump(results, args.out)
def main():
args = parse_args()
# load test results
outputs = mmcv.load(args.result)
assert ('pred_score' in outputs and 'pred_class' in outputs
and 'pred_label' in outputs), \
'No "pred_label", "pred_score" or "pred_class" in result file, ' \
'please set "--out-items" in test.py'
cfg = mmcv.Config.fromfile(args.config)
if args.cfg_options is not None:
cfg.merge_from_dict(args.cfg_options)
model = build_classifier(cfg.model)
# build the dataloader
dataset = build_dataset(cfg.data.test)
filenames = list()
for info in dataset.data_infos:
if info['img_prefix'] is not None:
filename = osp.join(info['img_prefix'],
info['img_info']['filename'])
else:
filename = info['img_info']['filename']
filenames.append(filename)
gt_labels = list(dataset.get_gt_labels())
gt_classes = [dataset.CLASSES[x] for x in gt_labels]
outputs['filename'] = filenames
outputs['gt_label'] = gt_labels
outputs['gt_class'] = gt_classes
outputs_list = list()
for i in range(len(gt_labels)):
output = dict()
for k in outputs.keys():
output[k] = outputs[k][i]
outputs_list.append(output)
# sort result
outputs_list = sorted(outputs_list, key=lambda x: x['pred_score'])
success = list()
fail = list()
for output in outputs_list:
if output['pred_label'] == output['gt_label']:
success.append(output)
else:
fail.append(output)
success = success[:args.topk]
fail = fail[:args.topk]
save_imgs(args.out_dir, 'success', success, model)
save_imgs(args.out_dir, 'fail', fail, model)
def test_repeat_dataset(self):
# Test build
dataset = build_dataset(
dict(type='RepeatDataset', dataset=self.dataset_cfg, times=3))
assert isinstance(dataset, RepeatDataset)
assert dataset.dataset.test_mode == self.dataset_cfg['test_mode']
# Test default_args
dataset = build_dataset(
dict(type='RepeatDataset', dataset=self.dataset_cfg, times=3),
{'test_mode': True})
assert dataset.dataset.test_mode == self.dataset_cfg['test_mode']
cp_cfg = deepcopy(self.dataset_cfg)
cp_cfg.pop('test_mode')
dataset = build_dataset(
dict(type='RepeatDataset', dataset=cp_cfg, times=3),
{'test_mode': True})
assert dataset.dataset.test_mode
def __init__(self, dataset_cfg, log_file_path, phase):
super(DatasetValidator, self).__init__()
# keep only LoadImageFromFile pipeline
assert dataset_cfg.data[phase].pipeline[0][
'type'] == 'LoadImageFromFile', 'This tool is only for dataset ' \
'that needs to load image from files.'
self.pipeline = PIPELINES.build(dataset_cfg.data[phase].pipeline[0])
dataset_cfg.data[phase].pipeline = []
dataset = build_dataset(dataset_cfg.data[phase])
self.dataset = dataset
self.log_file_path = log_file_path
def read_random_data(self):
train_dataset = build_dataset(self.cfg.data.train)
print(F"train sample number: {len(train_dataset)}")
train_dataloader = build_dataloader(
dataset=train_dataset,
samples_per_gpu=self.cfg.data.samples_per_gpu,
workers_per_gpu=self.cfg.data.workers_per_gpu,
num_gpus=1,
dist=False,
shuffle=False)
data = next(iter(train_dataloader))
return data
def test_create_dataloader(self):
train_dataset = build_dataset(self.cfg.data.train)
print(F"train sample number: {len(train_dataset)}")
train_dataloader = build_dataloader(
dataset=train_dataset,
samples_per_gpu=self.cfg.data.samples_per_gpu,
workers_per_gpu=self.cfg.data.workers_per_gpu,
num_gpus=2,
dist=False,
shuffle=False)
for index, data in enumerate(train_dataloader):
for k, v in data.items():
print(k, type(v.data[0]))
print(data["img"].data.shape)
print(data["gt_labels"].data)
print(data["gt_angles"].data)
break
def train_model(model,
dataset,
cfg,
distributed=False,
validate=False,
timestamp=None,
device=None,
meta=None):
logger = get_root_logger()
# prepare data loaders
dataset = dataset if isinstance(dataset, (list, tuple)) else [dataset]
sampler_cfg = cfg.data.get('sampler', None)
data_loaders = [
build_dataloader(
ds,
cfg.data.samples_per_gpu,
cfg.data.workers_per_gpu,
# cfg.gpus will be ignored if distributed
num_gpus=len(cfg.gpu_ids),
dist=distributed,
round_up=True,
seed=cfg.seed,
sampler_cfg=sampler_cfg) for ds in dataset
]
# put model on gpus
if distributed:
find_unused_parameters = cfg.get('find_unused_parameters', False)
# Sets the `find_unused_parameters` parameter in
# torch.nn.parallel.DistributedDataParallel
model = MMDistributedDataParallel(
model.cuda(),
device_ids=[torch.cuda.current_device()],
broadcast_buffers=False,
find_unused_parameters=find_unused_parameters)
else:
if device == 'cpu':
warnings.warn(
'The argument `device` is deprecated. To use cpu to train, '
'please refers to https://mmclassification.readthedocs.io/en'
'/latest/getting_started.html#train-a-model')
model = model.cpu()
else:
model = MMDataParallel(model, device_ids=cfg.gpu_ids)
if not model.device_ids:
from mmcv import __version__, digit_version
assert digit_version(__version__) >= (1, 4, 4), \
'To train with CPU, please confirm your mmcv version ' \
'is not lower than v1.4.4'
# build runner
optimizer = build_optimizer(model, cfg.optimizer)
if cfg.get('runner') is None:
cfg.runner = {
'type': 'EpochBasedRunner',
'max_epochs': cfg.total_epochs
}
warnings.warn(
'config is now expected to have a `runner` section, '
'please set `runner` in your config.', UserWarning)
runner = build_runner(cfg.runner,
default_args=dict(model=model,
batch_processor=None,
optimizer=optimizer,
work_dir=cfg.work_dir,
logger=logger,
meta=meta))
# an ugly walkaround to make the .log and .log.json filenames the same
runner.timestamp = timestamp
# fp16 setting
fp16_cfg = cfg.get('fp16', None)
if fp16_cfg is not None:
optimizer_config = Fp16OptimizerHook(**cfg.optimizer_config,
**fp16_cfg,
distributed=distributed)
elif distributed and 'type' not in cfg.optimizer_config:
optimizer_config = DistOptimizerHook(**cfg.optimizer_config)
else:
optimizer_config = cfg.optimizer_config
# register hooks
runner.register_training_hooks(cfg.lr_config,
optimizer_config,
cfg.checkpoint_config,
cfg.log_config,
cfg.get('momentum_config', None),
custom_hooks_config=cfg.get(
'custom_hooks', None))
if distributed and cfg.runner['type'] == 'EpochBasedRunner':
runner.register_hook(DistSamplerSeedHook())
# register eval hooks
if validate:
val_dataset = build_dataset(cfg.data.val, dict(test_mode=True))
val_dataloader = build_dataloader(
val_dataset,
samples_per_gpu=cfg.data.samples_per_gpu,
workers_per_gpu=cfg.data.workers_per_gpu,
dist=distributed,
shuffle=False,
round_up=True)
eval_cfg = cfg.get('evaluation', {})
eval_cfg['by_epoch'] = cfg.runner['type'] != 'IterBasedRunner'
eval_hook = DistEvalHook if distributed else EvalHook
# `EvalHook` needs to be executed after `IterTimerHook`.
# Otherwise, it will cause a bug if use `IterBasedRunner`.
# Refers to https://github.com/open-mmlab/mmcv/issues/1261
runner.register_hook(eval_hook(val_dataloader, **eval_cfg),
priority='LOW')
if cfg.resume_from:
runner.resume(cfg.resume_from)
elif cfg.load_from:
runner.load_checkpoint(cfg.load_from)
runner.run(data_loaders, cfg.workflow)
def main():
args = parse_args()
cfg = Config.fromfile(args.config)
if args.cfg_options is not None:
cfg.merge_from_dict(args.cfg_options)
# set multi-process settings
setup_multi_processes(cfg)
# set cudnn_benchmark
if cfg.get('cudnn_benchmark', False):
torch.backends.cudnn.benchmark = True
# work_dir is determined in this priority: CLI > segment in file > filename
if args.work_dir is not None:
# update configs according to CLI args if args.work_dir is not None
cfg.work_dir = args.work_dir
elif cfg.get('work_dir', None) is None:
# use config filename as default work_dir if cfg.work_dir is None
cfg.work_dir = osp.join('./work_dirs',
osp.splitext(osp.basename(args.config))[0])
if args.resume_from is not None:
cfg.resume_from = args.resume_from
if args.gpus is not None:
cfg.gpu_ids = range(1)
warnings.warn('`--gpus` is deprecated because we only support '
'single GPU mode in non-distributed training. '
'Use `gpus=1` now.')
if args.gpu_ids is not None:
cfg.gpu_ids = args.gpu_ids[0:1]
warnings.warn('`--gpu-ids` is deprecated, please use `--gpu-id`. '
'Because we only support single GPU mode in '
'non-distributed training. Use the first GPU '
'in `gpu_ids` now.')
if args.gpus is None and args.gpu_ids is None:
cfg.gpu_ids = [args.gpu_id]
# init distributed env first, since logger depends on the dist info.
if args.launcher == 'none':
distributed = False
else:
distributed = True
init_dist(args.launcher, **cfg.dist_params)
_, world_size = get_dist_info()
cfg.gpu_ids = range(world_size)
# create work_dir
mmcv.mkdir_or_exist(osp.abspath(cfg.work_dir))
# dump config
cfg.dump(osp.join(cfg.work_dir, osp.basename(args.config)))
# init the logger before other steps
timestamp = time.strftime('%Y%m%d_%H%M%S', time.localtime())
log_file = osp.join(cfg.work_dir, f'{timestamp}.log')
logger = get_root_logger(log_file=log_file, log_level=cfg.log_level)
# init the meta dict to record some important information such as
# environment info and seed, which will be logged
meta = dict()
# log env info
env_info_dict = collect_env()
env_info = '\n'.join([(f'{k}: {v}') for k, v in env_info_dict.items()])
dash_line = '-' * 60 + '\n'
logger.info('Environment info:\n' + dash_line + env_info + '\n' +
dash_line)
meta['env_info'] = env_info
# log some basic info
logger.info(f'Distributed training: {distributed}')
logger.info(f'Config:\n{cfg.pretty_text}')
# set random seeds
seed = init_random_seed(args.seed)
logger.info(f'Set random seed to {seed}, '
f'deterministic: {args.deterministic}')
set_random_seed(seed, deterministic=args.deterministic)
cfg.seed = seed
meta['seed'] = seed
model = build_classifier(cfg.model)
model.init_weights()
datasets = [build_dataset(cfg.data.train)]
if len(cfg.workflow) == 2:
val_dataset = copy.deepcopy(cfg.data.val)
val_dataset.pipeline = cfg.data.train.pipeline
datasets.append(build_dataset(val_dataset))
# save mmcls version, config file content and class names in
# runner as meta data
meta.update(
dict(mmcls_version=__version__,
config=cfg.pretty_text,
CLASSES=datasets[0].CLASSES))
# add an attribute for visualization convenience
train_model(model,
datasets,
cfg,
distributed=distributed,
validate=(not args.no_validate),
timestamp=timestamp,
device='cpu' if args.device == 'cpu' else 'cuda',
meta=meta)
def main():
args = parse_args()
cfg = mmcv.Config.fromfile(args.config)
# set cudnn_benchmark
if cfg.get('cudnn_benchmark', False):
torch.backends.cudnn.benchmark = True
cfg.model.pretrained = None
cfg.data.test.test_mode = True
# init distributed env first, since logger depends on the dist info.
if args.launcher == 'none':
distributed = False
else:
distributed = True
init_dist(args.launcher, **cfg.dist_params)
# build the dataloader
dataset = build_dataset(cfg.data.test)
data_loader = build_dataloader(dataset,
samples_per_gpu=cfg.data.samples_per_gpu,
workers_per_gpu=cfg.data.workers_per_gpu,
dist=distributed,
shuffle=False,
round_up=False)
# build the model and load checkpoint
model = build_classifier(cfg.model)
fp16_cfg = cfg.get('fp16', None)
if fp16_cfg is not None:
wrap_fp16_model(model)
_ = load_checkpoint(model, args.checkpoint, map_location='cpu')
if not distributed:
model = MMDataParallel(model, device_ids=[0])
outputs = single_gpu_test(model, data_loader)
else:
model = MMDistributedDataParallel(
model.cuda(),
device_ids=[torch.cuda.current_device()],
broadcast_buffers=False)
outputs = multi_gpu_test(model, data_loader, args.tmpdir,
args.gpu_collect)
rank, _ = get_dist_info()
if rank == 0:
nums = []
results = {}
for output in outputs:
nums.append(output['num_samples'].item())
for topk, v in output['accuracy'].items():
if topk not in results:
results[topk] = []
results[topk].append(v.item())
assert sum(nums) == len(dataset)
for topk, accs in results.items():
avg_acc = np.average(accs, weights=nums)
print(f'\n{topk} accuracy: {avg_acc:.2f}')
if args.out and rank == 0:
print(f'\nwriting results to {args.out}')
mmcv.dump(outputs, args.out)
def train_model(model,
dataset,
cfg,
distributed=False,
validate=False,
timestamp=None,
device='cuda',
meta=None):
logger = get_root_logger(cfg.log_level)
# prepare data loaders
dataset = dataset if isinstance(dataset, (list, tuple)) else [dataset]
data_loaders = [
build_dataloader(
ds,
cfg.data.samples_per_gpu,
cfg.data.workers_per_gpu,
# cfg.gpus will be ignored if distributed
num_gpus=len(cfg.gpu_ids),
dist=distributed,
round_up=True,
seed=cfg.seed) for ds in dataset
]
# put model on gpus
if distributed:
find_unused_parameters = cfg.get('find_unused_parameters', False)
# Sets the `find_unused_parameters` parameter in
# torch.nn.parallel.DistributedDataParallel
model = MMDistributedDataParallel(
model.cuda(),
device_ids=[torch.cuda.current_device()],
broadcast_buffers=False,
find_unused_parameters=find_unused_parameters)
else:
if device == 'cuda':
model = MMDataParallel(model.cuda(cfg.gpu_ids[0]),
device_ids=cfg.gpu_ids)
elif device == 'cpu':
model = model.cpu()
else:
raise ValueError(F'unsupported device name {device}.')
# build runner
optimizer = build_optimizer(model, cfg.optimizer)
if cfg.get('runner') is None:
cfg.runner = {
'type': 'EpochBasedRunner',
'max_epochs': cfg.total_epochs
}
warnings.warn(
'config is now expected to have a `runner` section, '
'please set `runner` in your config.', UserWarning)
runner = build_runner(cfg.runner,
default_args=dict(model=model,
batch_processor=None,
optimizer=optimizer,
work_dir=cfg.work_dir,
logger=logger,
meta=meta))
# an ugly walkaround to make the .log and .log.json filenames the same
runner.timestamp = timestamp
# fp16 setting
fp16_cfg = cfg.get('fp16', None)
if fp16_cfg is not None:
optimizer_config = Fp16OptimizerHook(**cfg.optimizer_config,
**fp16_cfg,
distributed=distributed)
elif distributed and 'type' not in cfg.optimizer_config:
optimizer_config = DistOptimizerHook(**cfg.optimizer_config)
else:
optimizer_config = cfg.optimizer_config
# register hooks
runner.register_training_hooks(cfg.lr_config,
optimizer_config,
cfg.checkpoint_config,
cfg.log_config,
cfg.get('momentum_config', None),
custom_hooks_config=cfg.get(
'custom_hooks', None))
if distributed:
runner.register_hook(DistSamplerSeedHook())
# register eval hooks
if validate:
val_dataset = build_dataset(cfg.data.val, dict(test_mode=True))
val_dataloader = build_dataloader(
val_dataset,
samples_per_gpu=cfg.data.samples_per_gpu,
workers_per_gpu=cfg.data.workers_per_gpu,
dist=distributed,
shuffle=False,
round_up=True)
eval_cfg = cfg.get('evaluation', {})
eval_cfg['by_epoch'] = cfg.runner['type'] != 'IterBasedRunner'
eval_hook = DistEvalHook if distributed else EvalHook
runner.register_hook(eval_hook(val_dataloader, **eval_cfg))
if cfg.resume_from:
runner.resume(cfg.resume_from)
elif cfg.load_from:
runner.load_checkpoint(cfg.load_from)
runner.run(data_loaders, cfg.workflow)
def train_model(model,
dataset,
cfg,
distributed=False,
validate=False,
timestamp=None,
meta=None):
logger = get_root_logger(cfg.log_level)
# prepare data loaders
dataset = dataset if isinstance(dataset, (list, tuple)) else [dataset]
data_loaders = [
build_dataloader(
ds,
cfg.data.samples_per_gpu,
cfg.data.workers_per_gpu,
# cfg.gpus will be ignored if distributed
num_gpus=len(cfg.gpu_ids),
dist=distributed,
round_up=True,
seed=cfg.seed) for ds in dataset
]
# put model on gpus
if distributed:
find_unused_parameters = cfg.get('find_unused_parameters', False)
# Sets the `find_unused_parameters` parameter in
# torch.nn.parallel.DistributedDataParallel
model = MMDistributedDataParallel(
model.cuda(),
device_ids=[torch.cuda.current_device()],
broadcast_buffers=False,
find_unused_parameters=find_unused_parameters)
else:
model = MMDataParallel(
model.cuda(cfg.gpu_ids[0]), device_ids=cfg.gpu_ids)
# build runner
optimizer = build_optimizer(model, cfg.optimizer)
runner = Runner(
model,
optimizer=optimizer,
work_dir=cfg.work_dir,
logger=logger,
meta=meta,
ema_cfg=cfg.get('ema_cfg', None))
# an ugly walkaround to make the .log and .log.json filenames the same
runner.timestamp = timestamp
# fp16 setting
fp16_cfg = cfg.get('fp16', None)
if fp16_cfg is not None:
optimizer_config = Fp16OptimizerHook(
**cfg.optimizer_config, **fp16_cfg, distributed=distributed)
elif distributed and 'type' not in cfg.optimizer_config:
optimizer_config = DistOptimizerHook(**cfg.optimizer_config)
else:
optimizer_config = cfg.optimizer_config
# register hooks
runner.register_training_hooks(cfg.lr_config, optimizer_config,
cfg.checkpoint_config, cfg.log_config,
cfg.get('momentum_config', None))
if distributed:
runner.register_hook(DistSamplerSeedHook())
# register eval hooks
if validate:
val_dataset = build_dataset(cfg.data.val, dict(test_mode=True))
val_dataloader = build_dataloader(
val_dataset,
samples_per_gpu=cfg.data.samples_per_gpu,
workers_per_gpu=cfg.data.workers_per_gpu,
dist=distributed,
shuffle=False,
round_up=False)
eval_cfg = cfg.get('evaluation', {})
eval_hook = DistEvalHook if distributed else EvalHook
runner.register_hook(eval_hook(val_dataloader, **eval_cfg))
if cfg.resume_from:
runner.resume(cfg.resume_from)
elif cfg.load_from:
runner.load_checkpoint(cfg.load_from)
runner.run(data_loaders, cfg.workflow, cfg.total_epochs)
def main():
args = parse_args()
cfg = mmcv.Config.fromfile(args.config)
if args.options is not None:
cfg.merge_from_dict(args.options)
# set cudnn_benchmark
if cfg.get('cudnn_benchmark', False):
torch.backends.cudnn.benchmark = True
cfg.model.pretrained = None
cfg.data.test.test_mode = True
# init distributed env first, since logger depends on the dist info.
if args.launcher == 'none':
distributed = False
else:
distributed = True
init_dist(args.launcher, **cfg.dist_params)
# build the dataloader
dataset = build_dataset(cfg.data.test)
data_loader = build_dataloader(dataset,
samples_per_gpu=cfg.data.samples_per_gpu,
workers_per_gpu=cfg.data.workers_per_gpu,
dist=distributed,
shuffle=False,
round_up=False)
# build the model and load checkpoint
model = build_classifier(cfg.model)
fp16_cfg = cfg.get('fp16', None)
if fp16_cfg is not None:
wrap_fp16_model(model)
checkpoint = load_checkpoint(model, args.checkpoint, map_location='cpu')
if not distributed:
model = MMDataParallel(model, device_ids=[0])
outputs = single_gpu_test(model, data_loader)
else:
model = MMDistributedDataParallel(
model.cuda(),
device_ids=[torch.cuda.current_device()],
broadcast_buffers=False)
outputs = multi_gpu_test(model, data_loader, args.tmpdir,
args.gpu_collect)
rank, _ = get_dist_info()
if rank == 0:
if args.metrics:
results = dataset.evaluate(outputs, args.metrics,
args.metric_options)
for k, v in results.items():
print(f'\n{k} : {v:.2f}')
else:
warnings.warn('Evaluation metrics are not specified.')
scores = np.vstack(outputs)
pred_score = np.max(scores, axis=1)
pred_label = np.argmax(scores, axis=1)
if 'CLASSES' in checkpoint['meta']:
CLASSES = checkpoint['meta']['CLASSES']
else:
from mmcls.datasets import ImageNet
warnings.simplefilter('once')
warnings.warn('Class names are not saved in the checkpoint\'s '
'meta data, use imagenet by default.')
CLASSES = ImageNet.CLASSES
pred_class = [CLASSES[lb] for lb in pred_label]
results = {
'pred_score': pred_score,
'pred_label': pred_label,
'pred_class': pred_class
}
if not args.out:
print('\nthe predicted result for the first element is '
f'pred_score = {pred_score[0]:.2f}, '
f'pred_label = {pred_label[0]} '
f'and pred_class = {pred_class[0]}. '
'Specify --out to save all results to files.')
if args.out and rank == 0:
print(f'\nwriting results to {args.out}')
mmcv.dump(results, args.out)
def main():
args = parse_args()
cfg = mmcv.Config.fromfile(args.config)
if args.cfg_options is not None:
cfg.merge_from_dict(args.cfg_options)
# set multi-process settings
setup_multi_processes(cfg)
# set cudnn_benchmark
if cfg.get('cudnn_benchmark', False):
torch.backends.cudnn.benchmark = True
cfg.model.pretrained = None
if args.gpu_ids is not None:
cfg.gpu_ids = args.gpu_ids[0:1]
warnings.warn('`--gpu-ids` is deprecated, please use `--gpu-id`. '
'Because we only support single GPU mode in '
'non-distributed testing. Use the first GPU '
'in `gpu_ids` now.')
else:
cfg.gpu_ids = [args.gpu_id]
# init distributed env first, since logger depends on the dist info.
if args.launcher == 'none':
distributed = False
else:
distributed = True
init_dist(args.launcher, **cfg.dist_params)
# build the dataloader
dataset = build_dataset(cfg.data.test, default_args=dict(test_mode=True))
# the extra round_up data will be removed during gpu/cpu collect
data_loader = build_dataloader(dataset,
samples_per_gpu=cfg.data.samples_per_gpu,
workers_per_gpu=cfg.data.workers_per_gpu,
dist=distributed,
shuffle=False,
round_up=True)
# build the model and load checkpoint
model = build_classifier(cfg.model)
fp16_cfg = cfg.get('fp16', None)
if fp16_cfg is not None:
wrap_fp16_model(model)
checkpoint = load_checkpoint(model, args.checkpoint, map_location='cpu')
if 'CLASSES' in checkpoint.get('meta', {}):
CLASSES = checkpoint['meta']['CLASSES']
else:
from mmcls.datasets import ImageNet
warnings.simplefilter('once')
warnings.warn('Class names are not saved in the checkpoint\'s '
'meta data, use imagenet by default.')
CLASSES = ImageNet.CLASSES
if not distributed:
if args.device == 'cpu':
model = model.cpu()
else:
model = MMDataParallel(model, device_ids=cfg.gpu_ids)
if not model.device_ids:
assert mmcv.digit_version(mmcv.__version__) >= (1, 4, 4), \
'To test with CPU, please confirm your mmcv version ' \
'is not lower than v1.4.4'
model.CLASSES = CLASSES
show_kwargs = {} if args.show_options is None else args.show_options
outputs = single_gpu_test(model, data_loader, args.show, args.show_dir,
**show_kwargs)
else:
model = MMDistributedDataParallel(
model.cuda(),
device_ids=[torch.cuda.current_device()],
broadcast_buffers=False)
outputs = multi_gpu_test(model, data_loader, args.tmpdir,
args.gpu_collect)
rank, _ = get_dist_info()
if rank == 0:
results = {}
logger = get_root_logger()
if args.metrics:
eval_results = dataset.evaluate(results=outputs,
metric=args.metrics,
metric_options=args.metric_options,
logger=logger)
results.update(eval_results)
for k, v in eval_results.items():
if isinstance(v, np.ndarray):
v = [round(out, 2) for out in v.tolist()]
elif isinstance(v, Number):
v = round(v, 2)
else:
raise ValueError(f'Unsupport metric type: {type(v)}')
print(f'\n{k} : {v}')
if args.out:
if 'none' not in args.out_items:
scores = np.vstack(outputs)
pred_score = np.max(scores, axis=1)
pred_label = np.argmax(scores, axis=1)
pred_class = [CLASSES[lb] for lb in pred_label]
res_items = {
'class_scores': scores,
'pred_score': pred_score,
'pred_label': pred_label,
'pred_class': pred_class
}
if 'all' in args.out_items:
results.update(res_items)
else:
for key in args.out_items:
results[key] = res_items[key]
print(f'\ndumping results to {args.out}')
mmcv.dump(results, args.out)
dict(type='ToTensor', keys=['img']),
dict(type='SpecCutout', keys=['img'], f_mask=8, t_mask=10),
dict(type='Collect', keys=['img', 'gt_label'])
]
test_pipeline = [
# dict(type='ToTensor', keys=['gt_label']),
dict(type='ToTensor', keys=['img']),
dict(type='Collect', keys=['img'])
]
data = dict(samples_per_gpu=128,
workers_per_gpu=2,
train=dict(type=dataset_type,
data_prefix='/data1/mayufeng/data_vox/train',
pipeline=train_pipeline,
map="/data1/mayufeng/data_vox/train/uid2classes.json"),
val=dict(type=dataset_type,
data_prefix='/data1/mayufeng/data_vox/val',
pipeline=test_pipeline,
map="/data1/mayufeng/data_vox/train/uid2classes.json"),
test=dict(type=dataset_type,
data_prefix='/data1/mayufeng/data_vox/val',
pipeline=test_pipeline,
map="/data1/mayufeng/data_vox/train/uid2classes.json"))
if __name__ == "__main__":
from mmcls.datasets import build_dataset
dataset = build_dataset(data["train"])
dataset.test_mode = True
print(dataset[1000])
def train_single_fold(args, cfg, fold, distributed, seed):
# create the work_dir for the fold
work_dir = osp.join(cfg.work_dir, f'fold{fold}')
cfg.work_dir = work_dir
# create work_dir
mmcv.mkdir_or_exist(osp.abspath(cfg.work_dir))
# wrap the dataset cfg
train_dataset = dict(
type='KFoldDataset',
fold=fold,
dataset=cfg.data.train,
num_splits=args.num_splits,
seed=seed,
)
val_dataset = dict(
type='KFoldDataset',
fold=fold,
# Use the same dataset with training.
dataset=copy.deepcopy(cfg.data.train),
num_splits=args.num_splits,
seed=seed,
test_mode=True,
)
val_dataset['dataset']['pipeline'] = cfg.data.val.pipeline
cfg.data.train = train_dataset
cfg.data.val = val_dataset
cfg.data.test = val_dataset
# dump config
stem, suffix = osp.basename(args.config).rsplit('.', 1)
cfg.dump(osp.join(cfg.work_dir, f'{stem}_fold{fold}.{suffix}'))
# init the logger before other steps
timestamp = time.strftime('%Y%m%d_%H%M%S', time.localtime())
log_file = osp.join(cfg.work_dir, f'{timestamp}.log')
logger = get_root_logger(log_file=log_file, log_level=cfg.log_level)
# init the meta dict to record some important information such as
# environment info and seed, which will be logged
meta = dict()
# log env info
env_info_dict = collect_env()
env_info = '\n'.join([(f'{k}: {v}') for k, v in env_info_dict.items()])
dash_line = '-' * 60 + '\n'
logger.info('Environment info:\n' + dash_line + env_info + '\n' +
dash_line)
meta['env_info'] = env_info
# log some basic info
logger.info(f'Distributed training: {distributed}')
logger.info(f'Config:\n{cfg.pretty_text}')
logger.info(
f'-------- Cross-validation: [{fold+1}/{args.num_splits}] -------- ')
# set random seeds
# Use different seed in different folds
logger.info(f'Set random seed to {seed + fold}, '
f'deterministic: {args.deterministic}')
set_random_seed(seed + fold, deterministic=args.deterministic)
cfg.seed = seed + fold
meta['seed'] = seed + fold
model = build_classifier(cfg.model)
model.init_weights()
datasets = [build_dataset(cfg.data.train)]
if len(cfg.workflow) == 2:
val_dataset = copy.deepcopy(cfg.data.val)
val_dataset.pipeline = cfg.data.train.pipeline
datasets.append(build_dataset(val_dataset))
meta.update(
dict(mmcls_version=__version__,
config=cfg.pretty_text,
CLASSES=datasets[0].CLASSES,
kfold=dict(fold=fold, num_splits=args.num_splits)))
# add an attribute for visualization convenience
train_model(model,
datasets,
cfg,
distributed=distributed,
validate=(not args.no_validate),
timestamp=timestamp,
device='cpu' if args.device == 'cpu' else 'cuda',
meta=meta)
def main():
args = parse_args()
cfg = mmcv.Config.fromfile(args.config)
if args.options is not None:
cfg.merge_from_dict(args.options)
# set cudnn_benchmark
if cfg.get('cudnn_benchmark', False):
torch.backends.cudnn.benchmark = True
cfg.model.pretrained = None
cfg.data.test.test_mode = True
assert args.metrics or args.out, \
'Please specify at least one of output path and evaluation metrics.'
# init distributed env first, since logger depends on the dist info.
if args.launcher == 'none':
distributed = False
else:
distributed = True
init_dist(args.launcher, **cfg.dist_params)
# build the dataloader
dataset = build_dataset(cfg.data.test)
# the extra round_up data will be removed during gpu/cpu collect
data_loader = build_dataloader(dataset,
samples_per_gpu=cfg.data.samples_per_gpu,
workers_per_gpu=cfg.data.workers_per_gpu,
dist=distributed,
shuffle=False,
round_up=True)
# build the model and load checkpoint
model = build_classifier(cfg.model)
fp16_cfg = cfg.get('fp16', None)
if fp16_cfg is not None:
wrap_fp16_model(model)
checkpoint = load_checkpoint(model, args.checkpoint, map_location='cpu')
if 'CLASSES' in checkpoint.get('meta', {}):
CLASSES = checkpoint['meta']['CLASSES']
else:
from mmcls.datasets import ImageNet
warnings.simplefilter('once')
warnings.warn('Class names are not saved in the checkpoint\'s '
'meta data, use imagenet by default.')
CLASSES = ImageNet.CLASSES
if not distributed:
if args.device == 'cpu':
model = model.cpu()
else:
model = MMDataParallel(model, device_ids=[0])
model.CLASSES = CLASSES
show_kwargs = {} if args.show_options is None else args.show_options
outputs = single_gpu_test(model, data_loader, args.show, args.show_dir,
**show_kwargs)
else:
model = MMDistributedDataParallel(
model.cuda(),
device_ids=[torch.cuda.current_device()],
broadcast_buffers=False)
outputs = multi_gpu_test(model, data_loader, args.tmpdir,
args.gpu_collect)
rank, _ = get_dist_info()
if rank == 0:
results = {}
if args.metrics:
eval_results = dataset.evaluate(outputs, args.metrics,
args.metric_options)
results.update(eval_results)
for k, v in eval_results.items():
print(f'\n{k} : {v:.2f}')
if args.out:
scores = np.vstack(outputs)
pred_score = np.max(scores, axis=1)
pred_label = np.argmax(scores, axis=1)
pred_class = [CLASSES[lb] for lb in pred_label]
results.update({
'class_scores': scores,
'pred_score': pred_score,
'pred_label': pred_label,
'pred_class': pred_class
})
print(f'\ndumping results to {args.out}')
mmcv.dump(results, args.out)
def main():
args = parse_args()
cfg = Config.fromfile(args.config)
if args.options is not None:
cfg.merge_from_dict(args.options)
# set cudnn_benchmark
if cfg.get('cudnn_benchmark', False):
torch.backends.cudnn.benchmark = True
if args.net_params:
tag, input_channels, block1, block2, block3, block4, last_channel = args.net_params.split('-')
input_channels = [int(item) for item in input_channels.split('_')]
block1 = [int(item) for item in block1.split('_')]
block2 = [int(item) for item in block2.split('_')]
block3 = [int(item) for item in block3.split('_')]
block4 = [int(item) for item in block4.split('_')]
last_channel = int(last_channel)
inverted_residual_setting = []
for item in [block1, block2, block3, block4]:
for _ in range(item[0]):
inverted_residual_setting.append([item[1], item[2:-int(len(item)/2-1)], item[-int(len(item)/2-1):]])
cfg.model.backbone.input_channel = input_channels
cfg.model.backbone.inverted_residual_setting = inverted_residual_setting
cfg.model.backbone.last_channel = last_channel
cfg.model.head.in_channels = last_channel
# work_dir is determined in this priority: CLI > segment in file > filename
if args.work_dir is not None:
# update configs according to CLI args if args.work_dir is not None
cfg.work_dir = args.work_dir
elif cfg.get('work_dir', None) is None:
# use config filename as default work_dir if cfg.work_dir is None
cfg.work_dir = osp.join('./work_dirs',
osp.splitext(osp.basename(args.config))[0])
if args.resume_from is not None:
cfg.resume_from = args.resume_from
if args.gpu_ids is not None:
cfg.gpu_ids = args.gpu_ids
else:
cfg.gpu_ids = range(1) if args.gpus is None else range(args.gpus)
if args.autoscale_lr:
# apply the linear scaling rule (https://arxiv.org/abs/1706.02677)
cfg.optimizer['lr'] = cfg.optimizer['lr'] * len(cfg.gpu_ids) / 8
# init distributed env first, since logger depends on the dist info.
if args.launcher == 'none':
distributed = False
else:
distributed = True
init_dist(args.launcher, **cfg.dist_params)
# create work_dir
mmcv.mkdir_or_exist(osp.abspath(cfg.work_dir))
# init the logger before other steps
timestamp = time.strftime('%Y%m%d_%H%M%S', time.localtime())
if args.net_params:
log_file = osp.join(cfg.work_dir, f'{args.net_params}.log')
else:
log_file = osp.join(cfg.work_dir, f'{timestamp}.log')
logger = get_root_logger(log_file=log_file, log_level=cfg.log_level)
# init the meta dict to record some important information such as
# environment info and seed, which will be logged
meta = dict()
# log env info
env_info_dict = collect_env()
env_info = '\n'.join([(f'{k}: {v}') for k, v in env_info_dict.items()])
dash_line = '-' * 60 + '\n'
logger.info('Environment info:\n' + dash_line + env_info + '\n' +
dash_line)
meta['env_info'] = env_info
# log some basic info
logger.info(f'Distributed training: {distributed}')
logger.info(f'Config:\n{cfg.pretty_text}')
# set random seeds
if args.seed is not None:
logger.info(f'Set random seed to {args.seed}, '
f'deterministic: {args.deterministic}')
set_random_seed(args.seed, deterministic=args.deterministic)
cfg.seed = args.seed
meta['seed'] = args.seed
model = build_classifier(cfg.model)
datasets = [build_dataset(cfg.data.train)]
if len(cfg.workflow) == 2:
val_dataset = copy.deepcopy(cfg.data.val)
val_dataset.pipeline = cfg.data.train.pipeline
datasets.append(build_dataset(val_dataset))
if cfg.checkpoint_config is not None:
# save mmcls version, config file content and class names in
# checkpoints as meta data
cfg.checkpoint_config.meta = dict(
mmcls_version=__version__,
config=cfg.pretty_text,
CLASSES=datasets[0].CLASSES)
# add an attribute for visualization convenience
train_model(
model,
datasets,
cfg,
distributed=distributed,
validate=(not args.no_validate),
timestamp=timestamp,
meta=meta)
def main():
print("--------> 1")
args = parse_args()
print("--------> 2")
cfg = Config.fromfile(args.config)
print("--------> 3")
if args.options is not None:
cfg.merge_from_dict(args.options)
# set cudnn_benchmark
if cfg.get('cudnn_benchmark', False):
torch.backends.cudnn.benchmark = True
print("--------> 4")
# work_dir is determined in this priority: CLI > segment in file > filename
if args.work_dir is not None:
# update configs according to CLI args if args.work_dir is not None
cfg.work_dir = args.work_dir
elif cfg.get('work_dir', None) is None:
# use config filename as default work_dir if cfg.work_dir is None
cfg.work_dir = osp.join('./work_dirs',
osp.splitext(osp.basename(args.config))[0])
if args.resume_from is not None:
cfg.resume_from = args.resume_from
if args.gpu_ids is not None:
cfg.gpu_ids = args.gpu_ids
else:
cfg.gpu_ids = range(1) if args.gpus is None else range(args.gpus)
print("--------> 5")
# init distributed env first, since logger depends on the dist info.
if args.launcher == 'none':
distributed = False
else:
distributed = True
init_dist(args.launcher, **cfg.dist_params)
_, world_size = get_dist_info()
cfg.gpu_ids = range(world_size)
print("--------> 6")
# create work_dir
mmcv.mkdir_or_exist(osp.abspath(cfg.work_dir))
# init the logger before other steps
timestamp = time.strftime('%Y%m%d_%H%M%S', time.localtime())
log_file = osp.join(cfg.work_dir, f'{timestamp}.log')
logger = get_root_logger(log_file=log_file, log_level=cfg.log_level)
print("--------> 7")
# init the meta dict to record some important information such as
# environment info and seed, which will be logged
meta = dict()
# log env info
env_info_dict = collect_env()
env_info = '\n'.join([(f'{k}: {v}') for k, v in env_info_dict.items()])
dash_line = '-' * 60 + '\n'
logger.info('Environment info:\n' + dash_line + env_info + '\n' +
dash_line)
meta['env_info'] = env_info
# log some basic info
cfg_pretty_text = cfg.pretty_text
logger.info(f'Distributed training: {distributed}')
logger.info(f'Config:\n{cfg_pretty_text}')
# set random seeds
if args.seed is not None:
logger.info(f'Set random seed to {args.seed}, '
f'deterministic: {args.deterministic}')
set_random_seed(args.seed, deterministic=args.deterministic)
cfg.seed = args.seed
meta['seed'] = args.seed
model = build_classifier(cfg.model)
datasets = [build_dataset(cfg.data.train)]
if len(cfg.workflow) == 2:
val_dataset = copy.deepcopy(cfg.data.val)
val_dataset.pipeline = cfg.data.train.pipeline
datasets.append(build_dataset(val_dataset))
if cfg.checkpoint_config is not None:
# save mmcls version, config file content and class names in
# checkpoints as meta data
cfg.checkpoint_config.meta = dict(mmcls_version=__version__,
config=cfg_pretty_text,
CLASSES=datasets[0].CLASSES)
model.CLASSES = datasets[0].CLASSES
# add an attribute for visualization convenience
train_model(model,
datasets,
cfg,
distributed=distributed,
validate=(not args.no_validate),
timestamp=timestamp,
device='cpu' if args.device == 'cpu' else 'cuda',
meta=meta)
def train_model(model,
dataset,
cfg,
distributed=False,
validate=False,
timestamp=None,
device=None,
meta=None):
"""Train a model.
This method will build dataloaders, wrap the model and build a runner
according to the provided config.
Args:
model (:obj:`torch.nn.Module`): The model to be run.
dataset (:obj:`mmcls.datasets.BaseDataset` | List[BaseDataset]):
The dataset used to train the model. It can be a single dataset,
or a list of dataset with the same length as workflow.
cfg (:obj:`mmcv.utils.Config`): The configs of the experiment.
distributed (bool): Whether to train the model in a distributed
environment. Defaults to False.
validate (bool): Whether to do validation with
:obj:`mmcv.runner.EvalHook`. Defaults to False.
timestamp (str, optional): The timestamp string to auto generate the
name of log files. Defaults to None.
device (str, optional): TODO
meta (dict, optional): A dict records some import information such as
environment info and seed, which will be logged in logger hook.
Defaults to None.
"""
logger = get_root_logger()
# prepare data loaders
dataset = dataset if isinstance(dataset, (list, tuple)) else [dataset]
# The default loader config
loader_cfg = dict(
# cfg.gpus will be ignored if distributed
num_gpus=cfg.ipu_replicas if device == 'ipu' else len(cfg.gpu_ids),
dist=distributed,
round_up=True,
seed=cfg.get('seed'),
sampler_cfg=cfg.get('sampler', None),
)
# The overall dataloader settings
loader_cfg.update({
k: v
for k, v in cfg.data.items() if k not in [
'train', 'val', 'test', 'train_dataloader', 'val_dataloader',
'test_dataloader'
]
})
# The specific dataloader settings
train_loader_cfg = {**loader_cfg, **cfg.data.get('train_dataloader', {})}
data_loaders = [build_dataloader(ds, **train_loader_cfg) for ds in dataset]
# put model on gpus
if distributed:
find_unused_parameters = cfg.get('find_unused_parameters', False)
# Sets the `find_unused_parameters` parameter in
# torch.nn.parallel.DistributedDataParallel
model = MMDistributedDataParallel(
model.cuda(),
device_ids=[torch.cuda.current_device()],
broadcast_buffers=False,
find_unused_parameters=find_unused_parameters)
else:
if device == 'cpu':
warnings.warn(
'The argument `device` is deprecated. To use cpu to train, '
'please refers to https://mmclassification.readthedocs.io/en'
'/latest/getting_started.html#train-a-model')
model = model.cpu()
elif device == 'ipu':
model = model.cpu()
else:
model = MMDataParallel(model, device_ids=cfg.gpu_ids)
if not model.device_ids:
from mmcv import __version__, digit_version
assert digit_version(__version__) >= (1, 4, 4), \
'To train with CPU, please confirm your mmcv version ' \
'is not lower than v1.4.4'
# build runner
optimizer = build_optimizer(model, cfg.optimizer)
if cfg.get('runner') is None:
cfg.runner = {
'type': 'EpochBasedRunner',
'max_epochs': cfg.total_epochs
}
warnings.warn(
'config is now expected to have a `runner` section, '
'please set `runner` in your config.', UserWarning)
if device == 'ipu':
if not cfg.runner['type'].startswith('IPU'):
cfg.runner['type'] = 'IPU' + cfg.runner['type']
if 'options_cfg' not in cfg.runner:
cfg.runner['options_cfg'] = {}
cfg.runner['options_cfg']['replicationFactor'] = cfg.ipu_replicas
cfg.runner['fp16_cfg'] = cfg.get('fp16', None)
runner = build_runner(cfg.runner,
default_args=dict(model=model,
batch_processor=None,
optimizer=optimizer,
work_dir=cfg.work_dir,
logger=logger,
meta=meta))
# an ugly walkaround to make the .log and .log.json filenames the same
runner.timestamp = timestamp
# fp16 setting
fp16_cfg = cfg.get('fp16', None)
if fp16_cfg is not None:
if device == 'ipu':
from mmcv.device.ipu import IPUFp16OptimizerHook
optimizer_config = IPUFp16OptimizerHook(
**cfg.optimizer_config,
loss_scale=fp16_cfg['loss_scale'],
distributed=distributed)
else:
optimizer_config = Fp16OptimizerHook(
**cfg.optimizer_config,
loss_scale=fp16_cfg['loss_scale'],
distributed=distributed)
elif distributed and 'type' not in cfg.optimizer_config:
optimizer_config = DistOptimizerHook(**cfg.optimizer_config)
else:
optimizer_config = cfg.optimizer_config
# register hooks
runner.register_training_hooks(cfg.lr_config,
optimizer_config,
cfg.checkpoint_config,
cfg.log_config,
cfg.get('momentum_config', None),
custom_hooks_config=cfg.get(
'custom_hooks', None))
if distributed and cfg.runner['type'] == 'EpochBasedRunner':
runner.register_hook(DistSamplerSeedHook())
# register eval hooks
if validate:
val_dataset = build_dataset(cfg.data.val, dict(test_mode=True))
# The specific dataloader settings
val_loader_cfg = {
**loader_cfg,
'shuffle': False, # Not shuffle by default
'sampler_cfg': None, # Not use sampler by default
**cfg.data.get('val_dataloader', {}),
}
val_dataloader = build_dataloader(val_dataset, **val_loader_cfg)
eval_cfg = cfg.get('evaluation', {})
eval_cfg['by_epoch'] = cfg.runner['type'] != 'IterBasedRunner'
eval_hook = DistEvalHook if distributed else EvalHook
# `EvalHook` needs to be executed after `IterTimerHook`.
# Otherwise, it will cause a bug if use `IterBasedRunner`.
# Refers to https://github.com/open-mmlab/mmcv/issues/1261
runner.register_hook(eval_hook(val_dataloader, **eval_cfg),
priority='LOW')
if cfg.resume_from:
runner.resume(cfg.resume_from)
elif cfg.load_from:
runner.load_checkpoint(cfg.load_from)
runner.run(data_loaders, cfg.workflow)
def main():
args = parse_args()
cfg = Config.fromfile(args.config)
if args.options is not None:
cfg.merge_from_dict(args.options)
# set cudnn_benchmark
if cfg.get('cudnn_benchmark', False):
torch.backends.cudnn.benchmark = True
if args.work_dir is not None:
# update configs according to CLI args if args.work_dir is not None
cfg.work_dir = args.work_dir
elif cfg.get('work_dir', None) is None:
# use config filename as default work_dir if cfg.work_dir is None
cfg.work_dir = osp.join('./work_dirs',
osp.splitext(osp.basename(args.config))[0])
if args.aml:
data_store = os.environ['AZUREML_DATAREFERENCE_{}'.format(
args.aml_data_store)]
parse(cfg, data_store)
if cfg.resume_from is not None:
cfg.resume_from = os.path.join(data_store,
args.aml_work_dir_prefix,
cfg.resume_from)
cfg.work_dir = os.path.join(data_store, args.aml_work_dir_prefix,
cfg.work_dir)
print('work_dir: ', cfg.work_dir)
if 'data' in cfg.model.pretrained:
cfg.model.pretrained = os.path.join(data_store,
cfg.model.pretrained)
# if not args.aml:
# # work_dir is determined in this priority: CLI > segment in file > filename
# if args.work_dir is not None:
# # update configs according to CLI args if args.work_dir is not None
# cfg.work_dir = args.work_dir
# elif cfg.get('work_dir', None) is None:
# # use config filename as default work_dir if cfg.work_dir is None
# cfg.work_dir = osp.join('./work_dirs',
# osp.splitext(osp.basename(args.config))[0])
if args.resume_from is not None:
cfg.resume_from = args.resume_from
if args.gpu_ids is not None:
cfg.gpu_ids = args.gpu_ids
else:
cfg.gpu_ids = range(1) if args.gpus is None else range(args.gpus)
if args.launcher == 'none':
distributed = False
else:
distributed = True
init_dist(args.launcher, **cfg.dist_params)
# update gpu num
if dist.is_initialized():
cfg.gpus = dist.get_world_size()
else:
cfg.gpus = args.gpus
if args.autoscale_lr:
# apply the linear scaling rule (https://arxiv.org/abs/1706.02677)
cfg.optimizer['lr'] = cfg.optimizer[
'lr'] * cfg.gpus / 8 * cfg.data.samples_per_gpu / 32
# create work_dir
mmcv.mkdir_or_exist(osp.abspath(cfg.work_dir))
# init the logger before other steps
timestamp = time.strftime('%Y%m%d_%H%M%S', time.localtime())
log_file = osp.join(cfg.work_dir, f'{timestamp}.log')
logger = get_root_logger(log_file=log_file, log_level=cfg.log_level)
# init the meta dict to record some important information such as
# environment info and seed, which will be logged
meta = dict()
# log env info
env_info_dict = collect_env()
env_info = '\n'.join([(f'{k}: {v}') for k, v in env_info_dict.items()])
dash_line = '-' * 60 + '\n'
logger.info('Environment info:\n' + dash_line + env_info + '\n' +
dash_line)
meta['env_info'] = env_info
# log some basic info
logger.info(f'Distributed training: {distributed}')
logger.info(f'Config:\n{cfg.pretty_text}')
# set random seeds
if args.seed is not None:
logger.info(f'Set random seed to {args.seed}, '
f'deterministic: {args.deterministic}')
set_random_seed(args.seed, deterministic=args.deterministic)
cfg.seed = args.seed
meta['seed'] = args.seed
model = build_classifier(cfg.model)
datasets = [build_dataset(cfg.data.train)]
if len(cfg.workflow) == 2:
val_dataset = copy.deepcopy(cfg.data.val)
val_dataset.pipeline = cfg.data.train.pipeline
datasets.append(build_dataset(val_dataset))
if cfg.checkpoint_config is not None:
# save mmcls version, config file content and class names in
# checkpoints as meta data
cfg.checkpoint_config.meta = dict(mmcls_version=__version__,
config=cfg.pretty_text,
CLASSES=datasets[0].CLASSES)
# add an attribute for visualization convenience
train_model(model,
datasets,
cfg,
distributed=distributed,
validate=(not args.no_validate),
timestamp=timestamp,
meta=meta)
def test_dataset_sample_number(self):
train_dataset = build_dataset(self.cfg.data.train)
print(F"train sample number: {len(train_dataset)}")
