'Places': '/home/zhmiao/datasets/Places365'}
parser = argparse.ArgumentParser()
parser.add_argument('--config', default='./config/Imagenet_LT/Stage_1.py', type=str)
parser.add_argument('--test', default=False, action='store_true')
parser.add_argument('--test_open', default=False, action='store_true')
parser.add_argument('--output_logits', default=False)
args = parser.parse_args()
test_mode = args.test
test_open = args.test_open
if test_open:
test_mode = True
output_logits = args.output_logits
config = source_import(args.config).config
training_opt = config['training_opt']
# change
relatin_opt = config['memory']
dataset = training_opt['dataset']
if not os.path.isdir(training_opt['log_dir']):
os.makedirs(training_opt['log_dir'])
print('Loading dataset from: %s' % data_root[dataset.rstrip('_LT')])
pprint.pprint(config)
if not test_mode:
sampler_defs = training_opt['sampler']
if sampler_defs:
parser.add_argument('--config',
default='./config/ImageNet_LT/stage_1.py',
type=str)
parser.add_argument('--test', default=False, action='store_true')
parser.add_argument('--test_open', default=False, action='store_true')
parser.add_argument('--output_logits', default=False)
parser.add_argument('--col', type=str)
args = parser.parse_args()
test_mode = args.test
test_open = args.test_open
if test_open:
test_mode = True
output_logits = args.output_logits
config = source_import(args.config).config
training_opt = config['training_opt']
# change
relatin_opt = config['memory']
dataset = training_opt['dataset']
data_loader = dataloader.load_data(data_root=data_root[dataset.rstrip('_LT')],
dataset=dataset,
phase='val',
batch_size=1,
num_workers=training_opt['num_workers'])
baseline_model = model(config, data_loader, test=False)
baseline_model.load_model()
for model in baseline_model.networks.values():
model.eval()
else:
data_root = data_root_dict[dataset.rstrip('_LT')]
print('Loading dataset from: %s' % data_root)
pprint.pprint(config)
# ============================================================================
# TRAINING
if not test_mode:
# during training, different sampler may be applied
sampler_defs = training_opt['sampler']
if sampler_defs:
if sampler_defs['type'] == 'ClassAwareSampler':
sampler_dic = {
'sampler':
source_import(sampler_defs['def_file']).get_sampler(),
'params': {
'num_samples_cls': sampler_defs['num_samples_cls']
}
}
elif sampler_defs['type'] in [
'MixedPrioritizedSampler', 'ClassPrioritySampler'
]:
sampler_dic = {
'sampler': source_import(sampler_defs['def_file']).get_sampler(),
'params': {k: v for k, v in sampler_defs.items() \
if k not in ['type', 'def_file']}
}
else:
sampler_dic = None