main.py

import argparse
from model import Model

parser = argparse.ArgumentParser(description="CLMVSNet args")

# network
parser.add_argument('--num_stage', type=int, default=3)
parser.add_argument('--base_channels', type=int, default=8)
parser.add_argument("--sample1", type=dict, nargs='+', default={"num_hypotheses":48, "interval_ratio":4})
parser.add_argument("--sample2", type=dict, nargs='+', default={"num_hypotheses":32, "interval_ratio":2})
parser.add_argument("--sample3", type=dict, nargs='+', default={"num_hypotheses":8, "interval_ratio":1})
parser.add_argument("--group", type=int, default=8)

# dataset
parser.add_argument("--img_size", type=int, nargs='+', default=[512, 640])
parser.add_argument("--datapath", default="/mnt/xkq/Data/MVS/train/dtu", type=str)
parser.add_argument("--trainlist", default="datasets/lists/dtu/train.txt", type=str)
parser.add_argument("--testlist", default="datasets/lists/dtu/test.txt", type=str)
parser.add_argument("--dataset_name", type=str, default="dtu_cl", choices=["dtu_cl", "general_eval"])
parser.add_argument('--batch_size', type=int, default=1, help='train batch size')
parser.add_argument('--numdepth', type=int, default=192, help='the number of depth values')
parser.add_argument('--interval_scale', type=float, default=1.06, help='the number of depth values')
parser.add_argument("--nviews", type=int, default=5)
parser.add_argument("--inverse_depth", action="store_true")

# training and val
parser.add_argument("--val", action="store_true")
parser.add_argument('--start_epoch', type=int, default=0)
parser.add_argument('--epochs', type=int, default=16, help='number of epochs to train')
parser.add_argument('--lr', type=float, default=0.0005, help='learning rate')
parser.add_argument('--wd', type=float, default=0.0, help='weight decay')
parser.add_argument('--scheduler', type=str, default="steplr", choices=["steplr", "cosinelr"])
parser.add_argument('--warmup', type=float, default=0.33, help='warmup epochs')
parser.add_argument('--milestones', type=float, nargs='+', default=[10, 12, 14], help='lr schedule')
parser.add_argument('--lr_decay', type=float, default=0.5, help='lr decay at every milestone')
parser.add_argument('--resume', type=str, help='path to the resume model')
parser.add_argument('--log_dir', type=str, default="./log", help='path to the log dir')
parser.add_argument('--dlossw', type=float, nargs='+', default=[0.5, 1.0, 2.0], help='depth loss weight for different stage')

# loss weights
parser.add_argument('--wrecon', type=float, default=8.0)
parser.add_argument('--w_icc', type=float, default=0.01)
parser.add_argument('--max_w_icc', type=float, default=0.32)
parser.add_argument('--w_scc', type=float, default=0.01)
parser.add_argument('--mask_conf', type=float, default=0.95)
parser.add_argument('--p_icc', type=float, default=0.1)

# log
parser.add_argument('--eval_freq', type=int, default=1, help='eval freq')
parser.add_argument('--summary_freq', type=int, default=50, help='print and summary frequency')

# testing
parser.add_argument("--test", action="store_true")
parser.add_argument('--outdir', default='./outputs', help='output dir')
parser.add_argument('--num_worker', type=int, default=4, help='depth_filer worker')
parser.add_argument('--filter_method', type=str, default='pcd', help="filter method")

# pcd
parser.add_argument('--conf', type=float, default=0.8, help='prob confidence, for pcd')
parser.add_argument('--thres_view', type=int, default=3, help='threshold of num view, for pcd')
parser.add_argument('--depth_thres', type=float, default=0.001, help='depth_thres for pcd')
parser.add_argument('--img_dist_thres', type=float, default=0.75, help='depth_thres for pcd')


# device and distributed
parser.add_argument("--no_cuda", action="store_true")
parser.add_argument("--local_rank", type=int, default=0)
parser.add_argument('--dist-url', default='env://', help='url used to set up distributed training')
parser.add_argument("--sync_bn", action="store_true")

args = parser.parse_args()

if __name__ == '__main__':
    model = Model(args)
    print(args)
    model.main()