if 'test' in cfg.data:
cfg.data.test.data_dir = args.data_dir
# 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)
# init logger before other steps
logger = get_root_logger(log_level=cfg.log_level)
# set random seeds
if args.seed is not None:
logger.info('Set random seed to {}'.format(args.seed))
set_random_seed(args.seed)
model = build_model(cfg.model)
fp16_cfg = cfg.get('fp16', None)
if fp16_cfg is not None:
wrap_fp16_model(model)
test_dataset = build_dataset(cfg.data.val)
test_network(model,
test_dataset,
cfg=cfg,
distributed=distributed,
logger=logger,
progress=args.progress)
else:
load_from = args.checkpoint
if load_from is None:
chkpt_list = [load_from]
elif os.path.isdir(load_from):
chkpt_list = [
os.path.join(load_from, fn) for fn in os.listdir(load_from)
if fn.endswith('.pth') and not fn.startswith('latest')
]
elif os.path.isfile(load_from):
chkpt_list = [load_from]
else:
raise FileNotFoundError("Cannot find {}".format(args.checkpoint))
cfg.work_dir = os.path.join(cfg.work_dir, 'clip_retrieval')
# build a dataloader
model = build_model(cfg.model,
default_args=dict(train_cfg=None, test_cfg=None))
train_dataset = build_dataset(cfg.data.train)
test_dataset = build_dataset(cfg.data.test)
results = test_clip_retrieval(model,
train_dataset,
test_dataset,
cfg,
chkpt_list,
logger=logger,
progress=args.progress)
init_dist(args.launcher, **cfg.dist_params)
# create work_dir
mmcv.mkdir_or_exist(os.path.abspath(cfg.work_dir))
# dump config
cfg.dump(os.path.join(cfg.work_dir, os.path.basename(args.cfg)))
# init logger before other steps
logger = get_root_logger(log_level=cfg.log_level)
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)
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('Set random seed to {}'.format(args.seed))
set_random_seed(args.seed)
model = build_model(cfg.model)
train_dataset = build_dataset(cfg.data.train)
train_network(model,
train_dataset,
cfg,
distributed=distributed,
validate=args.validate,
logger=logger)
args = parse_args()
cfg = Config.fromfile(args.cfg)
# update configs according to CLI args
cfg.gpus = args.gpus
cfg.data.videos_per_gpu = args.batchsize
if 'pretrained' in cfg['model']['backbone']:
cfg['model']['backbone']['pretrained'] = None
if args.work_dir is not None:
cfg.work_dir = args.work_dir
if args.data_dir is not None:
if 'test' in cfg.data:
cfg.data.test.root_dir = args.data_dir
if args.checkpoint is not None:
chkpt_list = [args.checkpoint]
else:
chkpt_list = [os.path.join(cfg.work_dir, fn)
for fn in os.listdir(cfg.work_dir) if fn.endswith('.pth')]
# init logger before other steps
logger = get_root_logger(log_level=cfg.log_level)
# build a dataloader
model = build_model(cfg.model, default_args=dict(train_cfg=cfg.train_cfg, test_cfg=cfg.test_cfg))
dataset = build_dataset(cfg.data.test)
results = test_recognizer(model,
dataset,
cfg,
chkpt_list,
logger=logger,
progress=args.progress)