-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathtest.py
160 lines (139 loc) · 5.39 KB
/
test.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
from torch.utils.data.dataloader import DataLoader
import torch
import torch.nn as nn
import torch.nn.functional as F
import numpy as np
import cv2
import matplotlib.pyplot as plt
from tqdm import tqdm
import os
import lib.image.warping as warping
from lib.model.localtrans import LocalTrans
from lib.data.dataset_homo import *
from lib.image.warping import *
from config.config import parse_config
class PSNR(nn.Module):
def __init__(self, max_val=1., mode='Y'):
super(PSNR, self).__init__()
self.max_val = max_val
self.mode = mode
def forward(self, x, y):
if self.mode == 'Y' and x.shape[1] == 3 and y.shape[1] == 3:
x = kornia.color.bgr_to_grayscale(x)
y = kornia.color.bgr_to_grayscale(y)
mse = F.mse_loss(x, y, reduction='mean')
psnr = 10 * torch.log10(self.max_val ** 2 / mse)
return psnr
class SSIM(nn.Module):
def __init__(self, window_size=11):
super(SSIM, self).__init__()
self.window_size = window_size
def forward(self, x, y):
if x.shape[1] == 3:
x = kornia.color.bgr_to_grayscale(x)
if y.shape[1] == 3:
y = kornia.color.bgr_to_grayscale(y)
return 1 - kornia.losses.ssim(x, y, self.window_size, 'mean')
def mkdir(dir):
try:
os.mkdir(dir)
except Exception:
pass
if __name__ == '__main__':
args = parse_config()
dataset = HomoDataset('/media/data1/shaoruizhi/AttentionWarping/val2014', bias=args.random_bias,
downsample=args.downsample, random_color=args.random_color,
random_noise=args.random_noise, random_identity=False)
dataloader = DataLoader(dataset, batch_size=args.batch_size, num_workers=4)
net1 = LocalTrans()
net2 = LocalTrans()
net3 = LocalTrans()
nets = [net1, net2, net3]
cuda = torch.device('cuda:%d' % args.gpu_id)
torch.cuda.set_device(args.gpu_id)
if args.resume or args.resume_dir is not None:
id = 0
if args.resume_dir is not None:
resume_dir = args.resume_dir
else:
resume_dir = os.path.join('results/checkpoints/%s' % args.name)
for net in nets:
if os.path.isfile(os.path.join(resume_dir, 'net%d_latest.pt' % id)):
print('load from %s' % (os.path.join(resume_dir, 'net%d_latest.pt' % id)))
net.load_state_dict(torch.load(os.path.join(resume_dir, 'net%d_latest.pt' % id)))
id += 1
for net in nets:
net.to(cuda)
net.eval()
np.random.seed(int(time.time()))
train_idx = 0
loss = 0
loss_list = {}
loss_list[0] = []
loss_list[1] = []
loss_list[2] = []
loss_list[3] = []
total_psnr = 0
total_ssim = 0
idx = 0
import time
total_time = 0
for data in tqdm(dataloader):
idx += 1
img1 = data["img1"].permute(0, 3, 1, 2).float().to(cuda)
o_img1 = img1.clone()
o_img2 = data["img2"].permute(0, 3, 1, 2).float().to(cuda)
hr_img = data["hr_img"].float().to(cuda)
img2 = o_img2.clone()
gt = data["gt"].reshape(-1, 2, 2, 2).to(cuda)
o_gt = data["gt"].reshape(-1, 2, 2, 2).to(cuda).clone()
result = []
t = time.time()
with torch.no_grad():
for level in range(0, 3):
net = nets[level]
flow = net(img1[:, :, 32:32+128, 32:32+128], img2[:, :, 32:32+128, 32:32+128], level)
result.append(flow.clone())
loss_level = torch.mean(torch.sqrt(torch.sum((flow.permute(0, 2, 3, 1) - gt)**2, dim=3)))
if level == 2:
loss += loss_level
dis = torch.sqrt(torch.sum((flow.permute(0, 2, 3, 1) - gt)**2, dim=3)).reshape(-1, 4)
dis = torch.mean(dis, dim=1)
loss_list[level].append(dis)
# if level != 2:
B = img1.shape[0]
grid = gen_grid(2, 2, 32, 128, 32, 128, B).to(cuda)
grid_flow = grid + flow.permute(0, 2, 3, 1)
homo = kornia.geometry.find_homography_dlt(grid_flow.reshape(B, -1, 2).contiguous(), grid.reshape(B, -1, 2).contiguous())
img2 = kornia.geometry.warp_perspective(img2, homo, (192, 192)).detach()
ori_grid = gen_grid(2, 2, 32, 128, 32, 128, B, device=cuda)
grid = ori_grid + gt
grid3 = torch.cat([grid.reshape(B, -1, 2), torch.ones(B, 4, 1, device=cuda)], dim=2)
grid3 = (homo @ grid3.transpose(1, 2)).transpose(1, 2)
grid2 = grid3[:, :, :2] / grid3[:, :, 2:]
gt = (grid2.reshape(B, 2, 2, 2) - ori_grid).detach()
print('level%d:' % level, loss_level)
# if level == 2:
# break
total_time += time.time() - t
train_idx += 1
for l in range(3):
loss_list[l] = torch.cat(loss_list[l], dim=0).detach().cpu().numpy()
fig, ax = plt.subplots()
ax.set_xscale("log")
ax.grid(True, 'both')
for l in range(3):
a = 1e-2
x = []
y = []
while True:
for i in range(1, 10):
x.append(a*i)
y.append(np.sum(loss_list[l] < a*i) / len(loss_list[l]))
a *= 10
if a >= 100:
break
print(y)
ax.plot(x, y)
plt.savefig('data.jpg')
print(loss)