distributed = int(os.environ['WORLD_SIZE']) > 1
if distributed:
torch.cuda.set_device(args.local_rank)
torch.distributed.init_process_group(backend='nccl',
init_method='env://')
dist_print(datetime.datetime.now().strftime('[%Y/%m/%d %H:%M:%S]') +
' start training...')
dist_print(cfg)
assert cfg.backbone in [
'18', '34', '50', '101', '152', '50next', '101next', '50wide',
'101wide'
]
train_loader, cls_num_per_lane = get_train_loader(
cfg.batch_size, cfg.data_root, cfg.griding_num, cfg.dataset,
cfg.use_aux, distributed, cfg.num_lanes)
net = parsingNet(pretrained=True,
backbone=cfg.backbone,
cls_dim=(cfg.griding_num + 1, cls_num_per_lane,
cfg.num_lanes),
use_aux=cfg.use_aux).cuda()
if distributed:
net = torch.nn.parallel.DistributedDataParallel(
net, device_ids=[args.local_rank])
optimizer = get_optimizer(net, cfg)
if cfg.finetune is not None:
dist_print('finetune from ', cfg.finetune)
work_dir = get_work_dir(cfg)
distributed = False
if 'WORLD_SIZE' in os.environ:
distributed = int(os.environ['WORLD_SIZE']) > 1
if distributed:
torch.cuda.set_device(args.local_rank)
torch.distributed.init_process_group(backend='nccl', init_method='env://')
dist_print(datetime.datetime.now().strftime('[%Y/%m/%d %H:%M:%S]') + ' start training...')
dist_print(cfg)
assert cfg.backbone in ['18','34','50','101','152','50next','101next','50wide','101wide']
train_loader, num_anchors = get_train_loader(cfg.batch_size, cfg.data_root, cfg.griding_num, cfg.dataset, cfg.use_aux, distributed, cfg.num_lanes, cfg.num_classes)
net = parsingNet(
pretrained=True,
backbone=cfg.backbone,
det_dim=(cfg.griding_num+1, num_anchors, cfg.num_lanes),
cls_dim=(cfg.num_classes, cfg.num_lanes),
use_aux=cfg.use_aux
).cuda()
if distributed:
net = torch.nn.parallel.DistributedDataParallel(net, device_ids = [args.local_rank])
optimizer = get_optimizer(net, cfg)
if cfg.finetune is not None:
dist_print('finetune from ', cfg.finetune)
torch.cuda.set_device(args.local_rank)
torch.distributed.init_process_group(backend='nccl',
init_method='env://')
w, h = cfg['dataset']['w'], cfg['dataset']['h']
net = parsingNet(network=cfg['network'], datasets=cfg['dataset']).cuda()
if distributed:
net = torch.nn.parallel.DistributedDataParallel(
net, device_ids=[args.local_rank])
# try:
# from thop import profile
# macs, params = profile(net, inputs=(torch.zeros(1, 3, h, w).to(device)))
# ms = 'FLOPs: %.2f GFLOPS, Params: %.2f M'%(params/ 1E9, params/ 1E6)
# except:
# ms = 'Model profile error'
# logger.log(ms)
train_loader = get_train_loader(cfg['dataset'], args.local_rank)
test_loader = get_test_loader(cfg['dataset'], args.local_rank)
optimizer = get_optimizer(net, cfg['train'])
if cfg['finetune'] is not None:
state_all = torch.load(cfg['finetune'])['model']
state_clip = {} # only use backbone parameters
for k, v in state_all.items():
if 'model' in k:
state_clip[k] = v
net.load_state_dict(state_clip, strict=False)
if cfg['resume'] is not None:
logger.log('==> Resume model from ' + cfg['resume'])
resume_dict = torch.load(cfg['resume'], map_location='cpu')
net.load_state_dict(resume_dict['model'])
if 'optimizer' in resume_dict.keys():
work_dir = get_work_dir(cfg)
distributed = False
if 'WORLD_SIZE' in os.environ:
distributed = int(os.environ['WORLD_SIZE']) > 1
if distributed:
torch.cuda.set_device(args.local_rank)
torch.distributed.init_process_group(backend='nccl',
init_method='env://')
dist_print(datetime.datetime.now().strftime('[%Y/%m/%d %H:%M:%S]') +
' start training...')
dist_print(cfg)
train_loader = get_train_loader(cfg.batch_size, cfg.data_root, cfg.dataset,
distributed)
net = E2ENet(Channels=96,
nums_lane=4,
culomn_channels=cfg.griding_num,
row_channels=cfg.row_num,
initialed=True)
net.to(device)
if distributed:
net = torch.nn.parallel.DistributedDataParallel(
net, device_ids=[args.local_rank])
if cfg.finetune is not None:
dist_print('finetune from ', cfg.finetune)