-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathtry_recon.py
141 lines (104 loc) · 6.07 KB
/
try_recon.py
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
import os
import argparse
import shutil
import torch
import torch.nn as nn
import torchvision
import torchvision.utils as vutils
import torch.backends.cudnn as cudnn
import tensorboardX
import numpy as np
from tqdm import tqdm
from torch.utils.data import DataLoader
import torchvision.utils as vutils
from Unet_util import *
from utils import prepare_sub_folder
from Unet import *
from DnCNN import *
parser = argparse.ArgumentParser()
parser.add_argument('--config', type=str, default='', help='Path to the config file.')
parser.add_argument('--output_path', type=str, default='.', help="outputs path")
img_path = '/data/bowen/SparseReconstruction/3d-ct-full-dose/'
# data_loader = get_data_loader("fbp_recons_test.npy", '/data/bowen/SparseReconstruction/3d-ct-full-dose/test_abdominal_imgs_ood_final.npy', batch_size = 16)
# data_loader = get_data_loader("nerp_recon_fbptest.npy", "nerp_recon_fbptest.npy",batch_size = 1)
# data_loader = get_data_loader("fbp_result_iter0.npy", "fbp_result_iter0.npy",batch_size = 1)
# data_loader = get_data_loader("test_smooth.npy", "test_smooth.npy", batch_size = 1)
# data_loader = get_data_loader(img_path + "combined_recon_iter1.npy", img_path + "combined_recon_iter1.npy", train = False, batch_size = 15)
# data_loader = get_data_loader(img_path+ 'fbp_imgs.npy', img_path+ 'fbp_imgs.npy', train = False, batch_size = 15)
# data_loader = get_data_loader("corrected_recon_iter1.npy", "corrected_recon_iter1.npy",batch_size = 1)
# data_loader = get_data_loader('rn_i0_nlx1.npy', 'rn_i0_nlx1.npy', batch_size = 1)
# data_loader = get_data_loader('cnntest.npy', 'cnntest.npy', batch_size = 1)
# data_loader = get_data_loader('fbp_test.npy', 'fbp_test.npy', train = False, batch_size = 15)
# data_loader = get_data_loader(img_path + 'fbp_cnn_recon.npy', img_path + 'fbp_cnn_recon.npy', train=False, batch_size=15)
# data_loader = get_data_loader(img_path + 'fbp_test_proj' + str(20) + '.npy', img_path + 'fbp_test_proj' + str(20) + '.npy', train = False, batch_size = 10)
# data_loader = get_data_loader('test_proj.npy', 'test_proj.npy', train = False, batch_size = 11) ###test adaptation to angles
# data_loader = get_data_loader('test_cnninput6.npy', 'test_cnninput6.npy', train = False, batch_size = 1)
# data_loader = get_data_loader('adapts_test.npy','adapts_test.npy', train = False, batch_size=10)
# data_loader = get_data_loader('adapts_test2.npy','adapts_test2.npy', train = False, batch_size=10)
# data_loader = get_data_loader('adapts_test3.npy','adapts_test3.npy', train = False, batch_size=10)
# data_loader = get_data_loader('adapts_test4.npy','adapts_test4.npy', train = False, batch_size=10)
# data_loader = get_data_loader("features_and_data/fbp_test.npy", "features_and_data/fbp_test.npy", train = False, batch_size=10) #####for reconstruction with the original fbp test set
data_loader = get_data_loader('adapts_prod.npy','adapts_prod.npy', train = False, batch_size=10) ####for production (input adaptation)
# data_loader = get_data_loader('cnn_adapt_prod.npy','cnn_adapt_prod.npy', train = False, batch_size=10)
# data_loader = get_data_loader('test_cnninput6_baseline.npy', 'test_cnninput6_baseline.npy', train = False, batch_size = 1)
# Load experiment setting
# data_loader = get_data_loader('features_and_data/fbp_test_complex_noise_unet.npy', 'features_and_data/fbp_test_complex_noise_unet.npy', train = False, batch_size = 10) ####for black-box model complex-noise recon
# data_loader = get_data_loader('features_and_data/fbp_test_lowdose_unet.npy', 'features_and_data/fbp_test_lowdose_unet.npy', train = False, batch_size = 10) ####for black-box model low-dose recon
opts = parser.parse_args()
config = get_config(opts.config)
max_iter = config['max_iter']
model_name = "Unet1"
# model_name = "Unet_8192022_robust"
# model_name = "Unet_8212022_robust"
# model_name = "DnCNN_6262022"
cudnn.benchmark = True
output_folder = '/data/bowen/SparseReconstruction/3d-ct-full-dose/models'
train_writer = tensorboardX.SummaryWriter(os.path.join(opts.output_path + "/logs", model_name))
output_directory = os.path.join(opts.output_path + "/outputs", model_name)
checkpoint_directory, image_directory = prepare_sub_folder(output_directory)
print(checkpoint_directory)
# # Setup model
model = Unet(use_dropout=True)
# model = Unet(use_dropout=True, num_down=3)
# model = DnCNN(channels = 1)
model_path = checkpoint_directory + '/model_000101.pt'
model.load_state_dict(torch.load(model_path)['net'])
model.eval()
model.cuda(2)
ret = np.zeros((300, 256, 256)) ###for black-box model reocn
# ret = np.zeros((100, 256,256))
# ret = np.zeros((1,256,256)) ###for single image recon
# ret = np.zeros((50,256,256)) ###for input adaptation
a,b = None, None
cur_head = 0
for it, (fbp_raw, gt) in enumerate(data_loader):
fbp_raw = fbp_raw.transpose(1,3).float()
gt = gt.transpose(1,3).float()
fbp_raw = fbp_raw.cuda(2)
gt = gt.cuda(2)
test_data = (fbp_raw, gt)
test_output = model(test_data[0])
test_img = test_output.transpose(1,3).float()
a = test_img.cpu().detach().numpy()
b = gt.cpu().detach().numpy()
ret[cur_head:cur_head+10, :,:] = a[:,:,:,0]
cur_head += 10
# np.save(img_path + "cnn_projected_iter2.npy", ret)
# np.save(img_path + 'fbp_cnn_recon.npy', ret)
# np.save(img_path + 'fbp_cnn_recon2.npy', ret)
# np.save(img_path + 'fbp_cnn_recon_proj20.npy', ret)
# np.save("test_prior.npy", ret[0])
# np.save("test_adaptation6.npy", ret[0])
# np.save('test_adaptation6_baseline.npy', ret[0])
# np.save('test_fbp_gt_iter2.npy', b)
# np.save('rn_cnn_projected_iter2.npy', a)
# np.save('cnn_projected2.npy',a)
# np.save("cnn_adapt.npy", ret)
# np.save("cnn_adapt2.npy", ret)
# np.save("cnn_adapt3.npy", ret)
# np.save("cnn_adapt4.npy", ret)
np.save("cnn_adapt_prod.npy", ret) ###for input adaptation
# np.save("features_and_data/unet_recons_complex_noise_unet.npy", ret) ###for recon complex noise
# np.save("features_and_data/unet_recons_lowdose_unet.npy", ret) ###for recon low dose noise
# np.save("features_and_data/unet_recons_robust_continuous.npy", ret) ###for robust recon
# np.save("features_and_data/DnCNN_recon_lowdose.npy", ret) ###for recon lowdose dncnn