KnowSAM/train_semi_SAM.py at main · taozh2017/KnowSAM · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
import argparse
import numpy as np
import random
import torch
import os
import logging
import sys
from tqdm import tqdm
from dataloader.dataset import build_Dataset
from torch.utils.data import DataLoader
from utils.utils import patients_to_slices
from dataloader.transforms import build_transforms, build_weak_strong_transforms
from dataloader.TwoStreamBatchSampler import TwoStreamBatchSampler

from trainer import Trainer


parser = argparse.ArgumentParser()
parser.add_argument('--data_path', type=str, default='./SampleData',
                    help='Name of Experiment')
parser.add_argument('--labeled_num', type=int, default=1,
                    help='Percentage of label quantity')

parser.add_argument('--dataset', type=str, default='/tumor_1',
                    help='Name of Experiment')
# parser.add_argument('--dataset', type=str, default='/ISIC_TrainDataset_10',
#                     help='Name of Experiment')
# parser.add_argument('--dataset', type=str, default='/thyroid_30',
#                     help='Name of Experiment')
# parser.add_argument('--dataset', type=str, default='/BrainMRI_30',
#                     help='Name of Experiment')

parser.add_argument('--num_classes', type=int,  default=2,
                    help='output channel of network')
parser.add_argument('--in_channels', type=int, default=3,
                    help='input channel of network')

parser.add_argument('-lr', type=float, default=1e-4, help='initial learning rate')
parser.add_argument('-UNet_lr', type=float, default=0.01, help='initial learning rate')
parser.add_argument('-VNet_lr', type=float, default=0.01, help='initial learning rate')
parser.add_argument('--image_size', type=int, default=256, help='image_size')
parser.add_argument('--point_nums', type=int, default=5, help='points number')
parser.add_argument('--box_nums', type=int, default=1, help='boxes number')
parser.add_argument('--mod', type=str, default='sam_adpt', help='mod type:seg,cls,val_ad')
parser.add_argument("--model_type", type=str, default="vit_b", help="sam model_type")
parser.add_argument('-thd', type=bool, default=False, help='3d or not')
parser.add_argument('--batch_size', type=int, default=24,
                    help='batch_size per gpu')
parser.add_argument('--labeled_bs', type=int, default=12,
                    help='labeled_batch_size per gpu')
parser.add_argument('--seed', type=int,  default=42,
                    help='random seed')

parser.add_argument('--mixed_iterations', type=int, default=12000,
                    help='maximum epoch number to train')
parser.add_argument('--max_iterations', type=int, default=50000,
                    help='maximum epoch number to train')

parser.add_argument('--n_fold', type=int, default=1,
                    help='maximum epoch number to train')
parser.add_argument('--consistency', type=float, default=0.1,
                    help='consistency')
parser.add_argument('--consistency_rampup', type=float,
                    default=200.0, help='consistency_rampup')
parser.add_argument('--device', type=str, default='cuda')
parser.add_argument("--multimask", type=bool, default=False, help="ouput multimask")
parser.add_argument("--encoder_adapter", type=bool, default=True, help="use adapter")
parser.add_argument("--sam_checkpoint", type=str, default="./sam_vit_b_01ec64.pth", help="sam checkpoint")

args = parser.parse_args()


def sigmoid_rampup(current, rampup_length):
    """Exponential rampup from https://arxiv.org/abs/1610.02242"""
    if rampup_length == 0:
        return 1.0
    else:
        current = np.clip(current, 0.0, rampup_length)
        phase = 1.0 - current / rampup_length
        return float(np.exp(-5.0 * phase * phase))


def worker_init_fn(worker_id):
    random.seed(args.seed + worker_id)


def get_current_consistency_weight(epoch):
    # Consistency ramp-up from https://arxiv.org/abs/1610.02242
    return args.consistency * sigmoid_rampup(epoch, args.consistency_rampup)


def train(args, snapshot_path):
    batch_size = args.batch_size
    max_iterations = args.max_iterations
    # model
    trainer = Trainer(args)
    # dataset
    data_transforms = build_weak_strong_transforms(args)
    train_dataset = build_Dataset(args=args, data_dir=args.data_path + args.dataset, split="train_semi",
                                  transform=data_transforms)
    val_dataset = build_Dataset(args=args, data_dir=args.data_path + args.dataset, split="val",
                                transform=data_transforms["valid_test"])

    # sampler
    total_slices = len(train_dataset)
    labeled_slice = patients_to_slices(args.dataset, args.labeled_num)
    labeled_idxs = list(range(0, labeled_slice))
    unlabeled_idxs = list(range(labeled_slice, total_slices))
    batch_sampler = TwoStreamBatchSampler(labeled_idxs, unlabeled_idxs, batch_size, batch_size-args.labeled_bs)

    # dataloader
    train_loader = DataLoader(train_dataset, batch_sampler=batch_sampler, num_workers=2, pin_memory=True, worker_init_fn=worker_init_fn)
    val_loader = DataLoader(val_dataset, batch_size=1, shuffle=False, num_workers=1)
    logging.info("{} iterations per epoch".format(len(train_loader)))
    max_epoch = max_iterations // len(train_loader) + 1
    iterator = tqdm(range(max_epoch), ncols=70)

    # else:
    iter_num = 0
    for _ in iterator:
        for i_batch, sampled_batch in enumerate(train_loader):
            volume_batch, label_batch = sampled_batch['image'].cuda(), sampled_batch['label'].cuda()
            trainer.train(volume_batch, label_batch, iter_num)
            iter_num = iter_num + 1
            if iter_num > 0 and iter_num % 200 == 0:
                if "ACDC" not in args.dataset:
                    trainer.val(val_loader, snapshot_path, iter_num)
                else:
                    trainer.val_ACDC(val_loader, snapshot_path, iter_num)


if __name__ == '__main__':
    import shutil
    for fold in range(args.n_fold):
        torch.autograd.set_detect_anomaly(True)
        random.seed(2024)
        np.random.seed(2024)
        torch.manual_seed(2024)
        torch.cuda.manual_seed(2024)

        snapshot_path = "./Results/Result_tumor_1/fold_" + str(fold)

        if not os.path.exists(snapshot_path):
            os.makedirs(snapshot_path)
        if os.path.exists(snapshot_path + '/code'):
            shutil.rmtree(snapshot_path + '/code')
        if not os.path.exists(snapshot_path + '/code'):
            os.makedirs(snapshot_path + '/code')

        shutil.copyfile("./train_semi_SAM.py", snapshot_path + "/code/train_semi_SAM.py")
        shutil.copyfile("./trainer.py", snapshot_path + "/code/trainer.py")

        logging.basicConfig(filename=snapshot_path+"/log.txt", level=logging.INFO,
                            format='[%(asctime)s.%(msecs)03d] %(message)s', datefmt='%H:%M:%S')
        logging.getLogger().addHandler(logging.StreamHandler(sys.stdout))
        logging.info(str(args))
        train(args, snapshot_path)