-
Notifications
You must be signed in to change notification settings - Fork 0
/
eval_functions.py
260 lines (194 loc) · 10.8 KB
/
eval_functions.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
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
import sys
sys.path.append('/local/tangytob/Summer2023/camera_calibration_synchronization/')
import numpy as np
import math
from xml.dom import minidom
import matplotlib.image as mpimg
import os
import json
import torch
from eval_human_pose import Metrics
metrics = Metrics()
def evaluate_relative(cam_intrinsics_array, cam_axis, cam_position, intrinsic_array_tsai, cam_axis_tsai, cam_position_tsai, output_path, data = 'result'):
results_focal_pred = []
results_focal_tsai = []
results_extriniscs_pred = []
results_extriniscs_tsai = []
results_position_diff = []
#print(cam_position)
for c in range(1, len(cam_position[0])):
results_position_diff.append(np.linalg.norm(cam_position[0][0] - cam_position[0][c] - (cam_position_tsai[0][0] - cam_position_tsai[0][c])))
focal_percentage = []
for c in range(len(cam_intrinsics_array)):
#print(len(intrinsic_array_tsai))
#print(len(cam_intrinsics_array), " hqwe123132")
cam_pred = cam_intrinsics_array[c]
cam_tsai = intrinsic_array_tsai[c]
#print(cam_pred, " cam pred")
#print(cam_tsai, " cam tsai")
results_focal_pred.append(cam_pred[0][0])
results_focal_tsai.append(cam_tsai[0][0])
focal_percentage.append(100*np.absolute(cam_pred[0][0] - cam_tsai[0][0])/cam_tsai[0][0])
rot_pred_ref = cam_axis[0]
rot_tsai_ref = cam_axis_tsai[0]
relative_angle_array = []
for c in range(1, len(cam_axis)):
rot_pred = cam_axis[c]
rot_tsai = cam_axis_tsai[c]
rotation_pred = np.linalg.inv(rot_pred_ref) @ rot_pred
rotation_tsai = np.linalg.inv(rot_tsai_ref) @ rot_tsai
relative_angle = np.rad2deg(np.linalg.inv(rotation_pred) @ rotation_tsai)
relative_angle1 = np.rad2deg(np.linalg.inv(rotation_tsai) @ rotation_pred)
print(relative_angle, relative_angle1, " ANGLE")
pred_angle = np.rad2deg(np.arccos((np.trace(rotation_pred)-1)/2))
tsai_angle = np.rad2deg(np.arccos((np.trace(rotation_tsai)-1)/2))
print("*************************************")
print(np.absolute(pred_angle - tsai_angle), pred_angle, tsai_angle, " PREDICTED DIFF, pred, tsai")
results_extriniscs_pred.append(pred_angle)
results_extriniscs_tsai.append(tsai_angle)
relative_angle_array.append(np.absolute(pred_angle - tsai_angle))
'''
#print(output_path + data, " output_path + data")
with open(output_path + data + '.csv','w') as file:
file.write('focal pred, focal tsai, angle difference, focal percent error, position error (meters)')
file.write('\n')
for i in range(len(results_focal_pred)):
if i == 0:
file.write(str(results_focal_pred[i]) + ',' + str(results_focal_tsai[i]) + ',' + str(0) + ',' + str(focal_percentage[i]) + ',' + str(0))
file.write('\n')
else:
file.write(str(results_focal_pred[i]) + ',' + str(results_focal_tsai[i]) + ',' + str(relative_angle_array[i - 1]) + ',' + str(focal_percentage[i]) + ',' + str(results_position_diff[i - 1]))
file.write('\n')
#return np.mean(results_focal_pred_array), np.mean(results_focal_tsai_array), np.mean(angle_diff_array), np.mean(error_npjpe_array), np.mean(focal_error_array), np.mean(results_position_diff_array)
print(results_position_diff, " result pos diff !!!!!!!!!!!!")
'''
return results_focal_pred, results_focal_tsai, np.mean(relative_angle_array), np.mean(focal_percentage), np.mean(results_position_diff)
def get_tsai(tsai_path, tsai_intrinsics_path, terrace_tsai_cal):
intrinsic_array = []
extrinsic_array = []
cam_position = []
cam_axis = []
for itr in range(len(terrace_tsai_cal)):
file = minidom.parse(tsai_path + terrace_tsai_cal[itr])
models_extrinsics = file.getElementsByTagName('Extrinsic')
models_instrinsics = file.getElementsByTagName('Intrinsic')
models_geometry = file.getElementsByTagName('Geometry')
tx = float(models_extrinsics[0].attributes.items()[0][1])
ty = float(models_extrinsics[0].attributes.items()[1][1])
tz = float(models_extrinsics[0].attributes.items()[2][1])
rx = float(models_extrinsics[0].attributes.items()[3][1])
ry = float(models_extrinsics[0].attributes.items()[4][1])
rz = float(models_extrinsics[0].attributes.items()[5][1])
img_x = float(models_geometry[0].attributes.items()[0][1])
img_y = float(models_geometry[0].attributes.items()[1][1])
ncx = float(models_geometry[0].attributes.items()[2][1])
nfx = float(models_geometry[0].attributes.items()[3][1])
dx = float(models_geometry[0].attributes.items()[4][1])
dy = float(models_geometry[0].attributes.items()[5][1])
dpx = float(models_geometry[0].attributes.items()[6][1])
dpy = float(models_geometry[0].attributes.items()[7][1])
#print(img_x, img_y, " gasaass")
f = float(models_instrinsics[0].attributes.items()[0][1])
kappa = float(models_instrinsics[0].attributes.items()[1][1])
t1 = float(models_instrinsics[0].attributes.items()[2][1])
t2 = float(models_instrinsics[0].attributes.items()[3][1])
sx = float(models_instrinsics[0].attributes.items()[4][1])
cam_matrix = np.array([[f, 0, t1 ],
[0, f, t2 ],
[0, 0, 1 ]
])
R_x = np.array([[1, 0, 0 ],
[0, math.cos(rx), -math.sin(rx) ],
[0, math.sin(rx), math.cos(rx) ]
])
R_y = np.array([[math.cos(ry), 0, math.sin(ry) ],
[0, 1, 0 ],
[-math.sin(ry), 0, math.cos(ry) ]
])
R_z = np.array([[math.cos(rz), -math.sin(rz), 0],
[math.sin(rz), math.cos(rz), 0],
[0, 0, 1]
])
R_rot = R_z @ R_y @ R_x
extrinsic_array.append(R_rot)
#cam_pos = (np.linalg.inv(R_rot)@np.array([tx,ty,tz]))
cam_pos = (np.linalg.inv(R_rot)@(-np.array([tx,ty,tz])))
cam_position.append(cam_pos)
#cam_axis.append([R_rot[:,0], R_rot[:,1], R_rot[:,2]])
cam_axis.append([np.linalg.inv(R_rot)[:,0], np.linalg.inv(R_rot)[:,1], np.linalg.inv(R_rot)[:,2]])
with open(tsai_intrinsics_path + "single_calibration_" + str(itr) + "_.json", 'r') as f:
cam_intrinsics = json.load(f)
intrinsic_array.append(cam_intrinsics["intrinsics"])
return intrinsic_array, extrinsic_array, cam_position, cam_axis
def evaluate(cam_intrinsics_array, cam_axis, cam_position, intrinsic_array_tsai, cam_axis_tsai, cam_position_tsai, output_path, data = 'result'):
results_focal_pred = []
results_focal_tsai = []
results_extriniscs_pred = []
results_extriniscs_tsai = []
results_position_diff = []
print(cam_position)
for c in range(len(cam_position[0])):
results_position_diff.append(np.linalg.norm(cam_position[0][c] - cam_position_tsai[0][c]))
for c in range(len(cam_intrinsics_array)):
#print(len(intrinsic_array_tsai))
#print(len(cam_intrinsics_array), " hqwe123132")
cam_pred = cam_intrinsics_array[c]
cam_tsai = intrinsic_array_tsai[c]
#print(cam_pred, " cam pred")
#print(cam_tsai, " cam tsai")
results_focal_pred.append(cam_pred[0][0])
results_focal_tsai.append(cam_tsai[0][0])
rot_pred_ref = cam_axis[0]
rot_tsai_ref = cam_axis_tsai[0]
for c in range(len(cam_axis)):
rot_pred = cam_axis[c]
rot_tsai = cam_axis_tsai[c]
rotation_pred = np.linalg.inv(rot_pred_ref) @ rot_pred
rotation_tsai = np.linalg.inv(rot_tsai_ref) @ rot_tsai
pred_angle = np.rad2deg(np.arccos((np.trace(rotation_pred)-1)/2))
tsai_angle = np.rad2deg(np.arccos((np.trace(rotation_tsai)-1)/2))
print(pred_angle, " pred angle")
print(tsai_angle, " tsai angle")
print(rotation_pred)
print(rotation_tsai)
print((np.trace(rotation_pred)-1)/2, " pred")
print((np.trace(rotation_tsai)-1)/2, " tsai")
results_extriniscs_pred.append(pred_angle)
results_extriniscs_tsai.append(tsai_angle)
#print(pred_angle, " rot_pred")
#print(tsai_angle, " rot_tsai")
angle_diff = []
for c in range(len(cam_axis)):
rot_pred = cam_axis[c]
rot_tsai = cam_axis_tsai[c]
rotation_pred = np.linalg.inv(rot_pred) @ rot_tsai
print((np.trace(rotation_pred)-1)/2, " THE ROTATION PRED TRACE", c)
pred_angle = np.rad2deg(np.arccos((np.trace(rotation_pred)-1)/2))
angle_diff.append(pred_angle)
print(pred_angle, " CAM " + str(c))
cam_position_torch = torch.tensor(cam_position).double()
cam_position_tsai_torch = torch.tensor(cam_position_tsai).double()
#print(cam_position_torch.shape, " HELOOAasssssssssss")
error_npjpe = metrics.mpjpe(cam_position_tsai_torch, cam_position_torch, use_scaling=True, root_joint=0)
print(error_npjpe)
results_focal_pred_array = []
results_focal_tsai_array = []
angle_diff_array = []
error_npjpe_array = []
focal_error_array = []
results_position_diff_array = []
with open(output_path + data + '.csv','w') as file:
file.write('focal pred, focal tsai, angle difference, nmpjpe, focal percent error, position error (meters)')
file.write('\n')
for i in range(len(results_focal_pred)):
focal_error = 100*np.absolute(results_focal_pred[i] - results_focal_tsai[i])/results_focal_tsai[i]
file.write(str(results_focal_pred[i]) + ',' + str(results_focal_tsai[i]) + ',' + str(angle_diff[i]) + ',' + str(error_npjpe.item()) + ',' + str(focal_error) + ',' + str(results_position_diff[i]))
file.write('\n')
results_focal_pred_array.append(results_focal_pred[i])
results_focal_tsai_array.append(results_focal_tsai[i])
angle_diff_array.append(angle_diff[i],)
error_npjpe_array.append( error_npjpe.item())
focal_error_array.append(focal_error)
results_position_diff_array.append(results_position_diff[i])
#results_focal_pred[i], results_focal_tsai[i], angle_diff[i], error_npjpe.item(), focal_error, results_position_diff[i]
return np.mean(results_focal_pred_array), np.mean(results_focal_tsai_array), np.mean(angle_diff_array), np.mean(error_npjpe_array), np.mean(focal_error_array), np.mean(results_position_diff_array)