def __init__(self, num_steps, device):
def weights_init(m):
if isinstance(m, nn.Conv1d):
init.kaiming_uniform_(m.weight)
if m.bias is not None:
init.zeros_(m.bias)
elif isinstance(m, nn.BatchNorm1d):
init.ones_(m.weight)
init.zeros_(m.bias)
network = Autoencoder(k=128, num_points=2048)
self.network = network.apply(weights_init).to(device)
self.loss = ChamferDistance()
self.optimizer = optim.Adam(self.network.parameters(),
lr=5e-4,
weight_decay=1e-5)
self.scheduler = CosineAnnealingLR(self.optimizer,
T_max=num_steps,
eta_min=1e-7)
parser.add_argument('--model', type=str, default=None)
parser.add_argument('--num_input', type=int, default=2048)
parser.add_argument('--num_coarse', type=int, default=1024)
parser.add_argument('--num_dense', type=int, default=16384)
parser.add_argument('--batch_size', type=int, default=16)
parser.add_argument('--alpha', type=float, default=0.1)
parser.add_argument('--loss_d1', type=str, default='cd')
parser.add_argument('--lr', type=float, default=1e-4)
parser.add_argument('--weight_decay', type=float, default=1e-6)
parser.add_argument('--epochs', type=int, default=100)
parser.add_argument('--num_workers', type=int, default=4)
args = parser.parse_args()
DEVICE = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')
cd_loss = ChamferDistance()
loss_d1 = cd_loss
train_dataset = ShapeNet(partial_path=args.partial_root, gt_path=args.gt_root, split='train')
val_dataset = ShapeNet(partial_path=args.partial_root, gt_path=args.gt_root, split='val')
train_dataloader = torch.utils.data.DataLoader(train_dataset, batch_size=args.batch_size, shuffle=True, num_workers=args.num_workers)
val_dataloader = torch.utils.data.DataLoader(val_dataset, batch_size=args.batch_size, shuffle=False, num_workers=args.num_workers)
network = Folding()
if args.model is not None:
print('Loaded trained model from {}.'.format(args.model))
network.load_state_dict(torch.load(args.model))
else:
print('Begin training new model.')
network.to(DEVICE)