-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathtanks.py
executable file
·188 lines (147 loc) · 6.85 KB
/
tanks.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
from torch.utils.data import Dataset
from datasets.data_io import *
import os
import numpy as np
import cv2
from collections import defaultdict
from PIL import Image
import torch
from torchvision import transforms as T
import math
class MVSDataset(Dataset):
def __init__(self, datapath, split='intermediate', n_views=3, img_wh=(1920, 1056)):
self.stages = 4
self.datapath = datapath
self.img_wh = img_wh
self.split = split
self.build_metas()
self.n_views = n_views
def build_metas(self):
self.metas = []
if self.split == 'intermediate':
self.scans = ['Family', 'Francis', 'Horse', 'Lighthouse',
'M60', 'Panther', 'Playground', 'Train']
self.image_sizes = {'Family': (1920, 1080),
'Francis': (1920, 1080),
'Horse': (1920, 1080),
'Lighthouse': (2048, 1080),
'M60': (2048, 1080),
'Panther': (2048, 1080),
'Playground': (1920, 1080),
'Train': (1920, 1080)}
elif self.split == 'advanced':
self.scans = ['Auditorium', 'Ballroom', 'Courtroom',
'Museum', 'Palace', 'Temple']
self.scans = ['Courtroom']
self.image_sizes = {'Auditorium': (1920, 1080),
'Ballroom': (1920, 1080),
'Courtroom': (1920, 1080),
'Museum': (1920, 1080),
'Palace': (1920, 1080),
'Temple': (1920, 1080)}
for scan in self.scans:
with open(os.path.join(self.datapath, self.split, scan, 'pair.txt')) as f:
num_viewpoint = int(f.readline())
for view_idx in range(num_viewpoint):
ref_view = int(f.readline().rstrip())
src_views = [int(x) for x in f.readline().rstrip().split()[1::2]]
if len(src_views) != 0:
self.metas += [(scan, -1, ref_view, src_views)]
def read_cam_file(self, filename):
with open(filename) as f:
lines = [line.rstrip() for line in f.readlines()]
# extrinsics: line [1,5), 4x4 matrix
extrinsics = np.fromstring(' '.join(lines[1:5]), dtype=np.float32, sep=' ')
extrinsics = extrinsics.reshape((4, 4))
# intrinsics: line [7-10), 3x3 matrix
intrinsics = np.fromstring(' '.join(lines[7:10]), dtype=np.float32, sep=' ')
intrinsics = intrinsics.reshape((3, 3))
depth_min = float(lines[11].split()[0])
depth_max = float(lines[11].split()[1])
return intrinsics, extrinsics, depth_min, depth_max
def read_img(self, filename, h, w):
img = Image.open(filename)
# scale 0~255 to 0~1
np_img = np.array(img, dtype=np.float32) / 255.
np_img = cv2.resize(np_img, self.img_wh, interpolation=cv2.INTER_LINEAR)
np_img_ms = {
"stage_3": cv2.resize(np_img, (w//8, h//8), interpolation=cv2.INTER_LINEAR),
"stage_2": cv2.resize(np_img, (w//4, h//4), interpolation=cv2.INTER_LINEAR),
"stage_1": cv2.resize(np_img, (w//2, h//2), interpolation=cv2.INTER_LINEAR),
"stage_0": np_img
}
return np_img_ms
def __len__(self):
return len(self.metas)
def __getitem__(self, idx):
scan, _, ref_view, src_views = self.metas[idx]
# use only the reference view and first nviews-1 source views
view_ids = [ref_view] + src_views[:self.n_views-1]
img_w, img_h = self.image_sizes[scan]
imgs_0 = []
imgs_1 = []
imgs_2 = []
imgs_3 = []
# depth = None
depth_min = None
depth_max = None
proj_matrices_0 = []
proj_matrices_1 = []
proj_matrices_2 = []
proj_matrices_3 = []
for i, vid in enumerate(view_ids):
img_filename = os.path.join(self.datapath, self.split, scan, f'images/{vid:08d}.jpg')
proj_mat_filename = os.path.join(self.datapath, self.split, scan, f'cams_1/{vid:08d}_cam.txt')
imgs = self.read_img(img_filename,self.img_wh[1], self.img_wh[0])
imgs_0.append(imgs['stage_0'])
imgs_1.append(imgs['stage_1'])
imgs_2.append(imgs['stage_2'])
imgs_3.append(imgs['stage_3'])
intrinsics, extrinsics, depth_min_, depth_max_ = self.read_cam_file(proj_mat_filename)
intrinsics[0] *= self.img_wh[0]/img_w
intrinsics[1] *= self.img_wh[1]/img_h
proj_mat = extrinsics.copy()
intrinsics[:2,:] *= 0.125
proj_mat[:3, :4] = np.matmul(intrinsics, proj_mat[:3, :4])
proj_matrices_3.append(proj_mat)
proj_mat = extrinsics.copy()
intrinsics[:2,:] *= 2
proj_mat[:3, :4] = np.matmul(intrinsics, proj_mat[:3, :4])
proj_matrices_2.append(proj_mat)
proj_mat = extrinsics.copy()
intrinsics[:2,:] *= 2
proj_mat[:3, :4] = np.matmul(intrinsics, proj_mat[:3, :4])
proj_matrices_1.append(proj_mat)
proj_mat = extrinsics.copy()
intrinsics[:2,:] *= 2
proj_mat[:3, :4] = np.matmul(intrinsics, proj_mat[:3, :4])
proj_matrices_0.append(proj_mat)
if i == 0: # reference view
depth_min = depth_min_
depth_max = depth_max_
# imgs: N*3*H0*W0, N is number of images
imgs_0 = np.stack(imgs_0).transpose([0, 3, 1, 2])
imgs_1 = np.stack(imgs_1).transpose([0, 3, 1, 2])
imgs_2 = np.stack(imgs_2).transpose([0, 3, 1, 2])
imgs_3 = np.stack(imgs_3).transpose([0, 3, 1, 2])
imgs = {}
imgs['stage_0'] = imgs_0
imgs['stage_1'] = imgs_1
imgs['stage_2'] = imgs_2
imgs['stage_3'] = imgs_3
# proj_matrices: N*4*4
proj_matrices_0 = np.stack(proj_matrices_0)
proj_matrices_1 = np.stack(proj_matrices_1)
proj_matrices_2 = np.stack(proj_matrices_2)
proj_matrices_3 = np.stack(proj_matrices_3)
proj={}
proj['stage_3']=proj_matrices_3
proj['stage_2']=proj_matrices_2
proj['stage_1']=proj_matrices_1
proj['stage_0']=proj_matrices_0
return {"imgs": imgs, # N*3*H0*W0
"proj_matrices": proj, # N*4*4
"depth_min": depth_min, # scalar
"depth_max": depth_max,
"filename": scan + '/{}/' + '{:0>8}'.format(view_ids[0]) + "{}"
}