-
Notifications
You must be signed in to change notification settings - Fork 7
/
warper.py
242 lines (194 loc) · 9.09 KB
/
warper.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
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
# Most from https://github.com/JiawangBian/SC-SfMLearner-Release/blob/master/inverse_warp.py
from __future__ import division
import torch
import torch.nn as nn
import torch.nn.functional as F
from scipy.spatial.transform import Rotation as R
import numpy as np
import cv2 as cv
import sys
class Warper(nn.Module):
def __init__(self, opt, intrinsic):
super(Warper, self).__init__()
self.height = opt.height
self.width = opt.width
self.depth_min = opt.min_depth
self.register_buffer('intrinsic', intrinsic.unsqueeze(0))
self.register_buffer('intrinsic_inv', intrinsic.double().inverse().float())
self.set_id_grid()
self.padding_mode = 'zeros'
def project_pixel(self, depth, pose):
bs = depth.size(0)
cam_coords = self.cam_coords * depth.unsqueeze(1)
proj_cam_to_src_pixel = self.intrinsic.expand(bs, 3, 3) @ pose[:, :3]
R = proj_cam_to_src_pixel[:, :, :3]
t = proj_cam_to_src_pixel[:, :, -1:]
src_pixel_coords, computed_depth, project_3d = self.cam2pixel(cam_coords, R, t)
valid_points = src_pixel_coords.abs().max(dim=-1)[0] <= 1
mask = valid_points.unsqueeze(1).float()
return src_pixel_coords, mask, computed_depth, cam_coords.squeeze(1), project_3d
def inverse_warp(self, img, depth, ref_depth, pose):
src_pixel_coords, mask, computed_depth, pt1, pt2 = self.project_pixel(depth, pose)
projected_img = F.grid_sample(img, src_pixel_coords,
padding_mode=self.padding_mode)
if ref_depth is not None:
projected_depth = F.grid_sample(ref_depth, src_pixel_coords,
padding_mode=self.padding_mode).clamp(min=self.depth_min)
else:
projected_depth = None
return projected_img, mask, projected_depth, computed_depth, src_pixel_coords, \
pt1, pt2
def inverse_flow(self, forward_flows):
# inverse optical flow: given a forward flow, return the backward flow
bs, h, w, _ = forward_flows.size()
x = forward_flows[..., 0].view(bs, -1)
y = forward_flows[..., 1].view(bs, -1)
l = torch.floor(x); r = l + 1
t = torch.floor(y); b = t + 1
mask = (l >= 0) * (t >= 0) * (r < w) * (b < h)
l *= mask; r *= mask; t *= mask; b *= mask
x *= mask; y *= mask
w_rb = torch.abs(x - l + 1e-3) * torch.abs(y - t + 1e-3)
w_rt = torch.abs(x - l + 1e-3) * torch.abs(b - y + 1e-3)
w_lb = torch.abs(r - x + 1e-3) * torch.abs(y - t + 1e-3)
w_lt = torch.abs(r - x + 1e-3) * torch.abs(b - y + 1e-3)
l = l.long(); r = r.long(); t = t.long(); b = b.long()
weight_maps = torch.zeros(bs, h, w).to(forward_flows.device).double()
grid_x = self.pixel_map[..., 0].view(-1).long()
grid_y = self.pixel_map[..., 1].view(-1).long()
for i in range(bs):
for j in self.idx_set:
weight_maps[i, t[i, j], l[i, j]] += w_lt[i, j]
weight_maps[i, t[i, j], r[i, j]] += w_rt[i, j]
weight_maps[i, b[i, j], l[i, j]] += w_lb[i, j]
weight_maps[i, b[i, j], r[i, j]] += w_rb[i, j]
forward_shifts = (-forward_flows + self.pixel_map.repeat(bs, 1, 1, 1)).double()
backward_flows = torch.zeros(forward_flows.size()).to(forward_shifts.device)
for i in range(bs):
for c in range(2):
for j in self.idx_set:
backward_flows[i, t[i, j], l[i, j], c] += \
forward_shifts[i, :, :, c].view(-1)[j] * w_lt[i, j]
backward_flows[i, t[i, j], r[i, j], c] += \
forward_shifts[i, :, :, c].view(-1)[j] * w_rt[i, j]
backward_flows[i, b[i, j], l[i, j], c] += \
forward_shifts[i, :, :, c].view(-1)[j] * w_lb[i, j]
backward_flows[i, b[i, j], r[i, j], c] += \
forward_shifts[i, :, :, c].view(-1)[j] * w_rb[i, j]
for c in range(2):
backward_flows[..., c] /= weight_maps
backward_flows[torch.isinf(backward_flows)] = 0
backward_flows[torch.isnan(backward_flows)] = 0
backward_flows += self.pixel_map.repeat(bs, 1, 1, 1)
backward_flows[weight_maps == 0] = -2
return backward_flows
def cam2pixel(self, cam_coords, R, t):
bs = cam_coords.size(0)
h, w = self.height, self.width
cam_coords_flat= cam_coords.reshape(bs, 3, -1)
pcoords = R @ cam_coords_flat + t
X = pcoords[:, 0]
Y = pcoords[:, 1]
Z = pcoords[:, 2].clamp(min=self.depth_min)
X_norm = 2*(X / Z) / (w - 1) - 1
Y_norm = 2*(Y / Z) / (h - 1) - 1
if self.padding_mode == 'zeros':
X_mask = ((X_norm > 1) + (X_norm < -1)).detach()
Y_mask = ((Y_norm > 1) + (Y_norm < -1)).detach()
pixel_coords = torch.stack([X_norm, Y_norm], dim=2) #[B, H*W, 2]
return pixel_coords.reshape(bs, h, w, 2), Z.reshape(bs, 1, h, w), \
pcoords.reshape(bs, 3, h, w)
def set_id_grid(self):
h, w = self.height , self.width
i_range = torch.arange(0, h).view(1, h, 1).expand(1, h, w).float()
j_range = torch.arange(0, w).view(1, 1, w).expand(1, h, w).float()
ones = torch.ones(1, h, w).float()
self.pixel_coords = torch.stack((j_range, i_range, ones), dim=1).reshape(1, 3, -1)
cam_coords = self.intrinsic_inv @ self.pixel_coords
cam_coords = cam_coords.reshape(1, 3, h, w)
self.register_buffer('cam_coords', cam_coords)
pixel_map = torch.cat((j_range, i_range), 0).unsqueeze(0)
self.register_buffer('pixel_map', pixel_map.permute(0, 2, 3, 1))
sw, sh = 3, 3
idxs = {i*sw+j: [] for i in range(sh) for j in range(sw)}
for i in range(h):
for j in range(w):
key = ((i % sh) * sw) + j % sw
idxs[key].append(i * w + j)
self.idx_set = [torch.Tensor(v).long() for v in idxs.values()]
def euler2mat(angle):
"""Convert euler angles to rotation matrix.
Reference: https://github.com/pulkitag/pycaffe-utils/blob/master/rot_utils.py#L174
Args:
angle: rotation angle along 3 axis (in radians) -- size = [B, 3]
Returns:
Rotation matrix corresponding to the euler angles -- size = [B, 3, 3]
"""
B = angle.size(0)
x, y, z = angle[:,0], angle[:,1], angle[:,2]
cosz = torch.cos(z)
sinz = torch.sin(z)
zeros = z.detach()*0
ones = zeros.detach()+1
zmat = torch.stack([cosz, -sinz, zeros,
sinz, cosz, zeros,
zeros, zeros, ones], dim=1).reshape(B, 3, 3)
cosy = torch.cos(y)
siny = torch.sin(y)
ymat = torch.stack([cosy, zeros, siny,
zeros, ones, zeros,
-siny, zeros, cosy], dim=1).reshape(B, 3, 3)
cosx = torch.cos(x)
sinx = torch.sin(x)
xmat = torch.stack([ones, zeros, zeros,
zeros, cosx, -sinx,
zeros, sinx, cosx], dim=1).reshape(B, 3, 3)
rotMat = xmat @ ymat @ zmat
return rotMat
def quat2mat(quat):
"""Convert quaternion coefficients to rotation matrix.
Args:
quat: first three coeff of quaternion of rotation. fourht is then computed to have a norm of 1 -- size = [B, 3]
Returns:
Rotation matrix corresponding to the quaternion -- size = [B, 3, 3]
"""
norm_quat = torch.cat([quat[:,:1].detach()*0 + 1, quat], dim=1)
norm_quat = norm_quat/norm_quat.norm(p=2, dim=1, keepdim=True)
w, x, y, z = norm_quat[:,0], norm_quat[:,1], norm_quat[:,2], norm_quat[:,3]
B = quat.size(0)
w2, x2, y2, z2 = w.pow(2), x.pow(2), y.pow(2), z.pow(2)
wx, wy, wz = w*x, w*y, w*z
xy, xz, yz = x*y, x*z, y*z
rotMat = torch.stack([w2 + x2 - y2 - z2, 2*xy - 2*wz, 2*wy + 2*xz,
2*wz + 2*xy, w2 - x2 + y2 - z2, 2*yz - 2*wx,
2*xz - 2*wy, 2*wx + 2*yz, w2 - x2 - y2 + z2], dim=1).reshape(B, 3, 3)
return rotMat
def pose_vec2mat(vec, rotation_mode='euler'):
"""
Convert 6DoF parameters to transformation matrix.
Args:s
vec: 6DoF parameters in the order of tx, ty, tz, rx, ry, rz -- [B, 6]
Returns:
A transformation matrix -- [B, 3, 4]
"""
translation = vec[:, :3].unsqueeze(-1) # [B, 3, 1]
rot = vec[:,3:]
if rotation_mode == 'euler':
rot_mat = euler2mat(rot) # [B, 3, 3]
elif rotation_mode == 'quat':
rot_mat = quat2mat(rot) # [B, 3, 3]
transform_mat = torch.cat([rot_mat, translation], dim=2) # [B, 3, 4]
bot = (transform_mat[:, -1, :].detach() * 0.).view(-1, 1, 4)
bot[:, :, -1] += 1.
transform_mat = torch.cat([transform_mat, bot], dim=1)
return transform_mat
def inverse_pose(pose_mat):
R = pose_mat[:, :3, :3]
t = pose_mat[:, :3, 3:]
R_T = torch.transpose(R, 1, 2)
t_inv = -R_T @ t
pose_inv = torch.cat([R_T, t_inv], dim=2)
bot = (pose_inv[:, -1, :].detach() * 0.).view(-1, 1, 4)
bot[:, :, -1] += 1.
pose_inv = torch.cat([pose_inv, bot], dim=1)
return pose_inv