-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathtest_net.py
190 lines (157 loc) · 7.63 KB
/
test_net.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
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
# coding: utf-8
import torch
import argparse
from torch.utils.data import DataLoader
import torchvision.transforms as transforms
import torch.backends.cudnn as cudnn
import numpy as np
import os
from torch.autograd import Variable
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import sys
sys.path.append(".")
from utils.utils import MyDataset, MyTestDataset
from tensorboardX import SummaryWriter
from datetime import datetime
# from net import MemNet as Net
from net import Generator as Generator
import cv2
import time
#----------------------------------------------------------------------------------
parser = argparse.ArgumentParser()
# parser.add_argument('--testfile', type=str, default='./data/sample_val_100.pkl', help='test image list') #sample_val_100 new_val_3k
# parser.add_argument('--test_path', type=str, default='./data/sampleval100/', help='test image file') #sampleval100 images_prev
parser.add_argument('--testfile', type=str, default='./data/lytro_pairs.pkl', help='test image list')
parser.add_argument('--test_path', type=str, default='./data/lytro', help='test image file')
# parser.add_argument('--testfile', type=str, default='./data/disp100.pkl', help='test image list')
# parser.add_argument('--test_path', type=str, default='./data/dispimages/', help='test image file')
parser.add_argument('--workers', type=int, default=2, help='number of data loading workers')
parser.add_argument('--cuda', action='store_true', default=True, help='enables cuda')
parser.add_argument('--imageSize', type=int, default=520, help='the low resolution image size')
parser.add_argument('--nGPU', type=int, default=2, help='number of GPUs to use')
parser.add_argument('--load_ckpt', type=str, default=None, help="path to finetune weights (to continue training)")
parser.add_argument('--out', type=str, default='./data/', help='output image dir')
parser.add_argument('--thread', type=float, default=0.5, help='thread for mask')
opt = parser.parse_args()
#-----------------------------output file-----------------------------------------------------------
# ./data/12-25_12-00-15_model_20/
result_dir = opt.out + opt.load_ckpt.split('/')[-2] + '_model_' + opt.load_ckpt.split('/')[-1].split('_')[0]+ '/'+ opt.test_path.split('/')[-1] +'/'
# result_dir = r'/data/guoxiaobao/Project/multi-focus-fusion/data/simul-convention/'
if not os.path.exists(result_dir):
os.makedirs(result_dir)
##------------------------check if cudnn and cuda are avaliable-----------
cudnn.benchmark = True
if torch.cuda.is_available() and not opt.cuda:
print("WARNING: You have a CUDA device, so you should probably run with --cuda")
# device = torch.device("cuda:0,1" if opt.cuda else "cpu")
ngpu = int(opt.nGPU)
# ------------------------------------ step 1/5 ------------------------------------
trainTransform = transforms.Compose([
transforms.ToTensor()
])
validTransform = transforms.Compose([
transforms.ToTensor()
])
test_data = MyTestDataset(path=opt.test_path,pkfile = opt.testfile, transform=trainTransform)
test_loader = DataLoader(dataset=test_data, batch_size=1, shuffle=False)
# ------------------------------------ step 2/5 ------------------------------------
n_res_blk = 12
net = Generator(n_res_blk)
print(net)
def _load_ckpt(model, ckpt):
load_dict = {}
for name in ckpt:
load_dict[name.split('module.')[1]]= ckpt[name]
model.load_state_dict(load_dict, strict=False)
return model
if opt.load_ckpt is not None:
# load params
pretrained_dict = torch.load(opt.load_ckpt)
net_state_dict = net.state_dict()
pretrained_dict_1 = {k: v for k, v in pretrained_dict.items() if k in net_state_dict}
net_state_dict.update(pretrained_dict_1)
net.load_state_dict(net_state_dict)
else: #train from scratch
raise Exception('none model!! set --load_ckpt')
if opt.cuda:
net = torch.nn.DataParallel(net)
net = net.cuda()
# -----------------------------step 3/5:define post processing---------------------------------
def postmerge(output1, output2, img1name, img2name,thread = 0.5):
im1 = cv2.imread(opt.test_path + '/' +img1name)
im2 = cv2.imread(opt.test_path + '/' + img2name)
# im1 = cv2.imread(result_dir + '/' +img1name)
# im2 = cv2.imread(result_dir + '/' + img2name)
output1 = (output1 + 1 -output2) / 2
#output1 = np.where(output1 >= thread, 1, 0).repeat(3, -1)#0210：non-binaray
#output1 = np.sign((output1 + 1 - output2) / 2)
output2 = 1 - output1
# output2 = np.where(output2 >= thread, 1, 0).repeat(3,-1)
#print(output1)
res = im1 * output1 + im2 * output2
return res
# def postmerge_single(output1, img1name, img2name,thread = 0.5):
# im1 = cv2.imread(opt.test_path + '/' +img1name)
# im2 = cv2.imread(opt.test_path + '/' + img2name)
#
# # output1 = (output1 + 1 -output2) / 2
# #output1 = np.where(output1 >= thread, 1, 0).repeat(3, -1)#0210：non-binaray
# #output1 = np.sign((output1 + 1 - output2) / 2)
# output2 = 1 - output1
# # output2 = np.where(output2 >= thread, 1, 0).repeat(3,-1)
# #print(output1)
# res = im1 * output1 + im2 * output2
# return res
def sliding_patch_preds(net, image, h_s, w_s, k_size):
#b, c, h, w = image.size()
# print(image.size())
b, c, h, w = image.size()
# print(image.size())
arr_cont = np.zeros((b,1,h,w),dtype=np.float32) # to record counts of patch for each pixel
arr_value = np.zeros((b,1,h,w),dtype=np.float32)
for x in range(0, h, h_s):
for y in range(0, w , w_s):
sub_img = image[:,:, x:np.clip((x + k_size),0,h), y:np.clip((y + k_size),0,w)] # slicing
sub_img = net(sub_img)
arr_value[:,:, x:np.clip((x + k_size),0,h), y:np.clip((y + k_size),0,w)] += sub_img.cpu().data.numpy()
arr_cont[:,:, x:np.clip((x + k_size),0,h), y:np.clip((y + k_size),0,w)] += 1
arr_value = arr_value/arr_cont
return arr_value
# ------------------------------------ step 4/5 --------------------------------------------------
net.eval()
for i, data in enumerate(test_loader):
image1, image2, img1name, img2name = data
image1, image2 = Variable(image1.cuda()), Variable(image2.cuda())
# print('image1size:',image1.size(),type(image1)) #image1size: torch.Size([1, 3, 520, 520]) <class 'torch.Tensor'>
mergeimg = torch.cat((image1,image2),1)
img1name = img1name[0]
img2name = img2name[0]
# print(img1name[0], img2name[0])
mergename = img1name.split('.')[0] + '_m.jpg' #id.jpg
print(mergename)
#tic = time.time()
output = net(mergeimg)
# print('mergeshape:',output.shape)
# print(img1name.shape)
# output = output.squeeze().cpu().data.numpy()[:,:,np.newaxis]
# result_img = postmerge_single(output, img1name, img2name, thread = opt.thread)
output1 = output[:,0,:,:].squeeze(0).cpu().data.numpy()[:,:,np.newaxis]
output2 = output[:,1,:,:].squeeze(0).cpu().data.numpy()[:,:,np.newaxis]
result_img = postmerge(output1, output2, img1name, img2name, thread = opt.thread)
# toc = time.time()
# print('time:',toc - tic)
# ------------------------------------ step5/5------------------------------------
# result_dir = r"./data/new_val_3k_result/"
# if not os.path.exists(result_dir):
# os.makedirs(result_dir)
cv2.imwrite(result_dir + '/' + mergename, result_img)
maskdir = result_dir + '/mask/'
if not os.path.exists(maskdir):
os.makedirs(maskdir)
# cv2.imwrite(maskdir + str(img1name), output.repeat(3,-1)*255) #single
cv2.imwrite(maskdir + str(img1name), output1.repeat(3,-1)*255)
cv2.imwrite(maskdir + str(img2name), output2.repeat(3,-1)*255)
cv2.imwrite(maskdir + str(mergename), (output1 + output2).repeat(3,-1)*255)
print('Finished testing')