def train(cfg, local_rank, distributed):
model = build_detection_model(cfg)
device = torch.device(cfg.MODEL.DEVICE)
model.to(device)
optimizer = make_optimizer(cfg, model)
scheduler = make_lr_scheduler(cfg, optimizer)
# Initialize mixed-precision training
use_mixed_precision = cfg.DTYPE == "float16"
amp_opt_level = 'O1' if use_mixed_precision else 'O0'
model, optimizer = amp.initialize(model,
optimizer,
opt_level=amp_opt_level)
if distributed:
model = torch.nn.parallel.DistributedDataParallel(
model,
device_ids=[local_rank],
output_device=local_rank,
# this should be removed if we update BatchNorm stats
broadcast_buffers=False,
)
arguments = {}
arguments["iteration"] = 0
output_dir = cfg.OUTPUT_DIR
save_to_disk = get_rank() == 0
checkpointer = DetectronCheckpointer(cfg, model, optimizer, scheduler,
output_dir, save_to_disk)
extra_checkpoint_data = checkpointer.load(cfg.MODEL.WEIGHT,
ignore=cfg.MODEL.VID.IGNORE)
if cfg.MODEL.VID.METHOD in ("fgfa", ):
checkpointer.load_flownet(cfg.MODEL.VID.FLOWNET_WEIGHT)
if not cfg.MODEL.VID.IGNORE:
arguments.update(extra_checkpoint_data)
data_loader = make_data_loader(
cfg,
is_train=True,
is_distributed=distributed,
start_iter=arguments["iteration"],
)
test_period = cfg.SOLVER.TEST_PERIOD
if test_period > 0:
data_loader_val = make_data_loader(cfg,
is_train=False,
is_distributed=distributed,
is_for_period=True)
else:
data_loader_val = None
checkpoint_period = cfg.SOLVER.CHECKPOINT_PERIOD
do_train(
cfg,
model,
data_loader,
data_loader_val,
optimizer,
scheduler,
checkpointer,
device,
checkpoint_period,
test_period,
arguments,
)
return model
def main():
parser = argparse.ArgumentParser(description="PyTorch Object Detection Inference")
parser.add_argument(
'--launcher',
choices=['pytorch', 'mpi'],
default='pytorch',
help='job launcher')
parser.add_argument(
"--config-file",
default="/private/home/fmassa/github/detectron.pytorch_v2/configs/e2e_faster_rcnn_R_50_C4_1x_caffe2.yaml",
metavar="FILE",
help="path to config file",
)
parser.add_argument("--local_rank", type=int, default=0)
parser.add_argument(
"--ckpt",
help="The path to the checkpoint for test, default is the latest checkpoint.",
default=None,
)
parser.add_argument(
"--motion-specific",
"-ms",
action="store_true",
help="if True, evaluate motion-specific iou"
)
parser.add_argument("--master_port", "-mp", type=str, default='29999')
parser.add_argument(
"opts",
help="Modify config options using the command-line",
default=None,
nargs=argparse.REMAINDER,
)
args = parser.parse_args()
if args.launcher == "pytorch":
num_gpus = int(os.environ["WORLD_SIZE"]) if "WORLD_SIZE" in os.environ else 1
elif args.launcher == "mpi":
num_gpus = int(os.environ["OMPI_COMM_WORLD_SIZE"]) if "OMPI_COMM_WORLD_SIZE" in os.environ else 1
else:
num_gpus = 1
distributed = num_gpus > 1
if distributed:
init_dist(args.launcher, args=args)
synchronize()
BASE_CONFIG = "configs/BASE_RCNN_{}gpu.yaml".format(num_gpus)
cfg.merge_from_file(BASE_CONFIG)
cfg.merge_from_file(args.config_file)
cfg.merge_from_list(args.opts)
cfg.freeze()
save_dir = ""
logger = setup_logger("mega_core", save_dir, get_rank())
logger.info("Using {} GPUs".format(num_gpus))
logger.info(cfg)
logger.info("Collecting env info (might take some time)")
logger.info("\n" + collect_env_info())
model = build_detection_model(cfg)
model.to(cfg.MODEL.DEVICE)
# Initialize mixed-precision if necessary
use_mixed_precision = cfg.DTYPE == 'float16'
amp_handle = amp.init(enabled=use_mixed_precision, verbose=cfg.AMP_VERBOSE)
output_dir = cfg.OUTPUT_DIR
checkpointer = DetectronCheckpointer(cfg, model, save_dir=output_dir)
ckpt = cfg.MODEL.WEIGHT if args.ckpt is None else args.ckpt
_ = checkpointer.load(ckpt, use_latest=args.ckpt is None, flownet=None)
iou_types = ("bbox",)
if cfg.MODEL.MASK_ON:
iou_types = iou_types + ("segm",)
if cfg.MODEL.KEYPOINT_ON:
iou_types = iou_types + ("keypoints",)
output_folders = [None] * len(cfg.DATASETS.TEST)
dataset_names = cfg.DATASETS.TEST
if cfg.OUTPUT_DIR:
for idx, dataset_name in enumerate(dataset_names):
output_folder = os.path.join(cfg.OUTPUT_DIR, "inference", dataset_name)
mkdir(output_folder)
output_folders[idx] = output_folder
data_loaders_val = make_data_loader(cfg, is_train=False, is_distributed=distributed)
for output_folder, dataset_name, data_loader_val in zip(output_folders, dataset_names, data_loaders_val):
inference(
cfg,
model,
data_loader_val,
dataset_name=dataset_name,
iou_types=iou_types,
motion_specific=args.motion_specific,
box_only=False if cfg.MODEL.RETINANET_ON else cfg.MODEL.RPN_ONLY,
bbox_aug=cfg.TEST.BBOX_AUG.ENABLED,
device=cfg.MODEL.DEVICE,
expected_results=cfg.TEST.EXPECTED_RESULTS,
expected_results_sigma_tol=cfg.TEST.EXPECTED_RESULTS_SIGMA_TOL,
output_folder=output_folder,
)
synchronize()
def main():
parser = argparse.ArgumentParser(
description="PyTorch Video Object Detection Training")
parser.add_argument(
"--config-file",
default="",
metavar="FILE",
help="path to config file",
type=str,
)
parser.add_argument('--launcher',
choices=['pytorch', 'mpi'],
default='pytorch',
help='job launcher')
parser.add_argument("--local_rank", type=int, default=0)
parser.add_argument(
"--skip-test",
dest="skip_test",
help="Do not test the final model",
action="store_true",
)
parser.add_argument("--master_port", "-mp", type=str, default='29999')
parser.add_argument("--save_name",
default="",
help="Where to store the log",
type=str)
parser.add_argument(
"opts",
help="Modify config options using the command-line",
default=None,
nargs=argparse.REMAINDER,
)
args = parser.parse_args()
num_gpus = int(
os.environ["WORLD_SIZE"]) if "WORLD_SIZE" in os.environ else 1
args.distributed = num_gpus > 1
if args.distributed:
init_dist(args.launcher, args=args)
synchronize()
# this is similar to the behavior of detectron2, which I think is a nice option.
BASE_CONFIG = "configs/BASE_RCNN_{}gpu.yaml".format(num_gpus)
cfg.merge_from_file(BASE_CONFIG)
cfg.merge_from_file(args.config_file)
cfg.merge_from_list(args.opts)
output_dir = cfg.OUTPUT_DIR
if output_dir:
mkdir(output_dir)
cfg.freeze()
logger = setup_logger("mega_core", output_dir, get_rank())
logger.info("Using {} GPUs".format(num_gpus))
logger.info(args)
logger.info("Collecting env info (might take some time)")
logger.info("\n" + collect_env_info())
logger.info("Loaded configuration file {}".format(args.config_file))
with open(args.config_file, "r") as cf:
config_str = "\n" + cf.read()
logger.info(config_str)
logger.info("Running with config:\n{}".format(cfg))
output_config_path = os.path.join(cfg.OUTPUT_DIR, 'config.yml')
logger.info("Saving config into: {}".format(output_config_path))
# save overloaded model config in the output directory
save_config(cfg, output_config_path)
if args.launcher == "mpi":
args.local_rank = ompi_local_rank()
model = train(cfg, args.local_rank, args.distributed)
if not args.skip_test:
run_test(cfg, model, args.distributed)