-
Notifications
You must be signed in to change notification settings - Fork 229
/
utils.py
executable file
·811 lines (683 loc) · 31.4 KB
/
utils.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
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
from __future__ import print_function
import torch
from torch.nn import functional as F
import numpy as np
import cv2, os, sys
import os.path as osp
from time import time
#from scipy.spatial.transform import Rotation as R
from threading import Thread
import re
#-----------------------------------------------------------------------------------------#
# IO utilizes
#-----------------------------------------------------------------------------------------#
def padding_image(image):
h, w = image.shape[:2]
side_length = max(h, w)
pad_image = np.zeros((side_length, side_length, 3), dtype=np.uint8)
top, left = int((side_length - h) // 2), int((side_length - w) // 2)
bottom, right = int(top+h), int(left+w)
pad_image[top:bottom, left:right] = image
image_pad_info = torch.Tensor([top, bottom, left, right, h, w])
return pad_image, image_pad_info
def img_preprocess(image, input_size=512):
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
pad_image, image_pad_info = padding_image(image)
input_image = torch.from_numpy(cv2.resize(pad_image, (input_size,input_size), interpolation=cv2.INTER_CUBIC))[None].float()
return input_image, image_pad_info
def convert_tensor2numpy(outputs, del_keys=['verts_camed','smpl_face', 'pj2d', 'verts_camed_org']):
for key in del_keys:
if key in outputs:
del outputs[key]
result_keys = list(outputs.keys())
for key in result_keys:
if isinstance(outputs[key], torch.Tensor):
outputs[key] = outputs[key].cpu().numpy()
return outputs
class ResultSaver:
def __init__(self, mode='image', save_path=None, save_npz=True):
self.is_dir = len(osp.splitext(save_path)[1]) == 0
self.mode = mode
self.save_path = save_path
self.save_npz = save_npz
self.save_dir = save_path if self.is_dir else osp.dirname(save_path)
if self.mode in ['image', 'video']:
os.makedirs(self.save_dir, exist_ok=True)
if self.mode == 'video':
self.frame_save_paths = []
def __call__(self, outputs, input_path, prefix=None, img_ext='.png'):
if self.mode == 'video' or self.is_dir:
save_name = osp.basename(input_path)
save_path = osp.join(self.save_dir, osp.splitext(save_name)[0])+img_ext
elif self.mode == 'image':
save_path = self.save_path
if prefix is not None:
save_path = osp.splitext(save_path)[0]+f'_{prefix}'+osp.splitext(save_path)[1]
rendered_image = None
if outputs is not None:
if 'rendered_image' in outputs:
rendered_image = outputs.pop('rendered_image')
if self.save_npz:
np.savez(osp.splitext(save_path)[0]+'.npz', results=outputs)
if rendered_image is None:
rendered_image = cv2.imread(input_path)
cv2.imwrite(save_path, rendered_image)
if self.mode == 'video':
self.frame_save_paths.append(save_path)
def save_video(self, save_path, frame_rate=24):
if len(self.frame_save_paths)== 0:
return
height, width = cv2.imread(self.frame_save_paths[0]).shape[:2]
writer = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*'mp4v'), frame_rate, (width, height))
for frame_path in self.frame_save_paths:
writer.write(cv2.imread(frame_path))
writer.release()
def save_video_results(frame_save_paths):
video_results = {}
video_sequence_results = {}
for frame_id, save_path in enumerate(frame_save_paths):
npz_path = osp.splitext(save_path)[0]+'.npz'
frame_results = np.load(npz_path, allow_pickle=True)['results'][()]
base_name = osp.basename(save_path)
video_results[base_name] = frame_results
if 'track_ids' not in frame_results:
continue
for subj_ind, track_id in enumerate(frame_results['track_ids']):
if track_id not in video_sequence_results:
video_sequence_results[track_id] = {'frame_id':[]}
video_sequence_results[track_id]['frame_id'].append(frame_id)
for key in frame_results:
if key not in video_sequence_results[track_id]:
video_sequence_results[track_id][key] = []
video_sequence_results[track_id][key].append(frame_results[key][subj_ind])
video_results_save_path = osp.join(osp.dirname(frame_save_paths[0]), 'video_results.npz')
np.savez(video_results_save_path, results=video_results, sequence_results=video_sequence_results)
class WebcamVideoStream(object):
def __init__(self, src=0):
# initialize the video camera stream and read the first frame
# from the stream
try:
self.stream = cv2.VideoCapture(src)
except:
self.stream = cv2.VideoCapture("/dev/video{}".format(src), cv2.CAP_V4L2)
(self.grabbed, self.frame) = self.stream.read()
# initialize the variable used to indicate if the thread should
# be stopped
self.stopped = False
def start(self):
# start the thread to read frames from the video stream
Thread(target=self.update, args=()).start()
return self
def update(self):
# keep looping infinitely until the thread is stopped
while True:
# if the thread indicator variable is set, stop the thread
if self.stopped:
return
# otherwise, read the next frame from the stream
(self.grabbed, self.frame) = self.stream.read()
def read(self):
# return the frame most recently read
return self.frame
def stop(self):
# indicate that the thread should be stopped
self.stopped = True
def video2frame(video_path, frame_save_dir=None):
cap = cv2.VideoCapture(video_path)
for frame_id in range(int(cap.get(cv2.CAP_PROP_FRAME_COUNT))):
success_flag, frame = cap.read()
if success_flag:
save_path = os.path.join(frame_save_dir, '{:08d}.jpg'.format(frame_id))
cv2.imwrite(save_path, frame)
def collect_frame_path(video_path, save_path):
assert osp.exists(video_path), video_path + 'not exist!'
is_dir = len(osp.splitext(save_path)[1]) == 0
if is_dir:
save_dir = save_path
save_name = osp.splitext(osp.basename(video_path))[0] + '.mp4'
else:
save_dir = osp.dirname(save_path)
save_name = osp.splitext(osp.basename(save_path))[0] + '.mp4'
video_save_path = osp.join(save_dir, save_name)
if osp.isfile(video_path):
video_name, video_ext = osp.splitext(osp.basename(video_path))
frame_save_dir = osp.join(save_dir, video_name+'_frames')
print(f'Extracting the frames of input {video_path} to {frame_save_dir}')
os.makedirs(frame_save_dir, exist_ok=True)
try:
video2frame(video_path, frame_save_dir)
except:
raise Exception(f"Failed in extracting the frames of {video_path} to {frame_save_dir}! \
Please check the video. If you want to do this by yourself, please extracte frames to {frame_save_dir} and take it as input to ROMP. \
For example, the first frame name is supposed to be {osp.join(frame_save_dir, '00000000.jpg')}")
else:
frame_save_dir = video_path
assert osp.isdir(frame_save_dir), frame_save_dir + 'is supposed to be a folder containing video frames.'
frame_paths = [osp.join(frame_save_dir, frame_name) for frame_name in sorted(os.listdir(frame_save_dir))]
return frame_paths, video_save_path
#-----------------------------------------------------------------------------------------#
# tracking & temporal optimization utils
#-----------------------------------------------------------------------------------------#
def smooth_global_rot_matrix(pred_rots, OE_filter):
rot_mat = batch_rodrigues(pred_rots[None]).squeeze(0)
smoothed_rot_mat = OE_filter.process(rot_mat)
smoothed_rot = rotation_matrix_to_angle_axis(smoothed_rot_mat.reshape(1,3,3)).reshape(-1)
return smoothed_rot
device = pred_rots.device
#print('before',pred_rots)
rot_euler = transform_rot_representation(pred_rots.cpu().numpy(), input_type='vec',out_type='mat')
smoothed_rot = OE_filter.process(rot_euler)
smoothed_rot = transform_rot_representation(smoothed_rot, input_type='mat',out_type='vec')
smoothed_rot = torch.from_numpy(smoothed_rot).float().to(device)
#print('after',smoothed_rot)
return smoothed_rot
class LowPassFilter:
def __init__(self):
self.prev_raw_value = None
self.prev_filtered_value = None
def process(self, value, alpha):
if self.prev_raw_value is None:
s = value
else:
s = alpha * value + (1.0 - alpha) * self.prev_filtered_value
self.prev_raw_value = value
self.prev_filtered_value = s
return s
class OneEuroFilter:
def __init__(self, mincutoff=1.0, beta=0.0, dcutoff=1.0, freq=30):
# min_cutoff: Decreasing the minimum cutoff frequency decreases slow speed jitter
# beta: Increasing the speed coefficient(beta) decreases speed lag.
self.freq = freq
self.mincutoff = mincutoff
self.beta = beta
self.dcutoff = dcutoff
self.x_filter = LowPassFilter()
self.dx_filter = LowPassFilter()
def compute_alpha(self, cutoff):
te = 1.0 / self.freq
tau = 1.0 / (2 * np.pi * cutoff)
return 1.0 / (1.0 + tau / te)
def process(self, x, print_inter=False):
prev_x = self.x_filter.prev_raw_value
dx = 0.0 if prev_x is None else (x - prev_x) * self.freq
edx = self.dx_filter.process(dx, self.compute_alpha(self.dcutoff))
if isinstance(edx, float):
cutoff = self.mincutoff + self.beta * np.abs(edx)
elif isinstance(edx, np.ndarray):
cutoff = self.mincutoff + self.beta * np.abs(edx)
elif isinstance(edx, torch.Tensor):
cutoff = self.mincutoff + self.beta * torch.abs(edx)
if print_inter:
print(self.compute_alpha(cutoff))
return self.x_filter.process(x, self.compute_alpha(cutoff))
def check_filter_state(OE_filters, signal_ID, show_largest=False, smooth_coeff=3.):
if len(OE_filters)>100:
del OE_filters
if signal_ID not in OE_filters:
if show_largest:
OE_filters[signal_ID] = create_OneEuroFilter(smooth_coeff)
else:
OE_filters[signal_ID] = {}
if len(OE_filters[signal_ID])>1000:
del OE_filters[signal_ID]
def create_OneEuroFilter(smooth_coeff):
return {'smpl_thetas': OneEuroFilter(smooth_coeff, 0.7), 'cam': OneEuroFilter(1.6, 0.7), 'smpl_betas': OneEuroFilter(0.6, 0.7), 'global_rot': OneEuroFilter(smooth_coeff, 0.7)}
def smooth_results(filters, body_pose=None, body_shape=None, cam=None):
if body_pose is not None:
global_rot = smooth_global_rot_matrix(body_pose[:3], filters['global_rot'])
body_pose = torch.cat([global_rot, filters['smpl_thetas'].process(body_pose[3:])], 0)
if body_shape is not None:
body_shape = filters['smpl_betas'].process(body_shape)
if cam is not None:
cam = filters['cam'].process(cam)
return body_pose, body_shape, cam
def euclidean_distance(detection, tracked_object):
return np.linalg.norm(detection.points - tracked_object.estimate)
def get_tracked_ids(detections, tracked_objects):
tracked_ids_out = np.array([obj.id for obj in tracked_objects])
tracked_points = np.array([obj.last_detection.points for obj in tracked_objects])
org_points = np.array([obj.points for obj in detections])
tracked_ids = [tracked_ids_out[np.argmin(np.linalg.norm(tracked_points-point[None], axis=1))] for point in org_points]
return tracked_ids
def get_tracked_ids3D(detections, tracked_objects):
tracked_ids_out = np.array([obj.id for obj in tracked_objects])
tracked_points = np.array([obj.last_detection.points for obj in tracked_objects])
org_points = np.array([obj.points for obj in detections])
tracked_ids = [tracked_ids_out[np.argmin(np.linalg.norm(tracked_points.reshape(-1,4)-point.reshape(1,4), axis=1))] for point in org_points]
return tracked_ids
#-----------------------------------------------------------------------------------------#
# 3D-to-2D projection utils
#-----------------------------------------------------------------------------------------#
INVALID_TRANS=np.ones(3)*-1
def convert_kp2d_from_input_to_orgimg(kp2ds, offsets):
offsets = offsets.float().to(kp2ds.device)
img_pad_size, crop_trbl, pad_trbl = offsets[:,:2], offsets[:,2:6], offsets[:,6:10]
leftTop = torch.stack([crop_trbl[:,3]-pad_trbl[:,3], crop_trbl[:,0]-pad_trbl[:,0]],1)
kp2ds_on_orgimg = (kp2ds + 1) * img_pad_size.unsqueeze(1) / 2 + leftTop.unsqueeze(1)
return kp2ds_on_orgimg
def convert_cam_to_3d_trans(cams, weight=2.):
(s, tx, ty) = cams[:,0], cams[:,1], cams[:,2]
depth, dx, dy = 1./s, tx/s, ty/s
trans3d = torch.stack([dx, dy, depth], 1)*weight
return trans3d
def batch_orth_proj(X, camera, mode='2d',keep_dim=False):
camera = camera.view(-1, 1, 3)
X_camed = X[:,:,:2] * camera[:, :, 0].unsqueeze(-1)
X_camed += camera[:, :, 1:]
if keep_dim:
X_camed = torch.cat([X_camed, X[:,:,2].unsqueeze(-1)],-1)
return X_camed
def vertices_kp3d_projection(outputs, meta_data=None, presp=False):
vertices, j3ds = outputs['verts'], outputs['j3d']
verts_camed = batch_orth_proj(vertices, outputs['cam'], mode='3d',keep_dim=True)
pj3d = batch_orth_proj(j3ds, outputs['cam'], mode='2d')
predicts_j3ds = j3ds[:,:24].contiguous().detach().cpu().numpy()
predicts_pj2ds = (pj3d[:,:,:2][:,:24].detach().cpu().numpy()+1)*256
cam_trans = estimate_translation(predicts_j3ds, predicts_pj2ds, \
focal_length=443.4, img_size=np.array([512,512])).to(vertices.device)
projected_outputs = {'verts_camed': verts_camed, 'pj2d': pj3d[:,:,:2], 'cam_trans':cam_trans}
if meta_data is not None:
projected_outputs['pj2d_org'] = convert_kp2d_from_input_to_orgimg(projected_outputs['pj2d'], meta_data['offsets'])
return projected_outputs
def estimate_translation_cv2(joints_3d, joints_2d, focal_length=600, img_size=np.array([512.,512.]), proj_mat=None, cam_dist=None):
if proj_mat is None:
camK = np.eye(3)
camK[0,0], camK[1,1] = focal_length, focal_length
camK[:2,2] = img_size//2
else:
camK = proj_mat
ret, rvec, tvec,inliers = cv2.solvePnPRansac(joints_3d, joints_2d, camK, cam_dist,\
flags=cv2.SOLVEPNP_EPNP,reprojectionError=20,iterationsCount=100)
if inliers is None:
return INVALID_TRANS
else:
tra_pred = tvec[:,0]
return tra_pred
def estimate_translation_np(joints_3d, joints_2d, joints_conf, focal_length=600, img_size=np.array([512.,512.]), proj_mat=None):
"""Find camera translation that brings 3D joints joints_3d closest to 2D the corresponding joints_2d.
Input:
joints_3d: (25, 3) 3D joint locations
joints: (25, 3) 2D joint locations and confidence
Returns:
(3,) camera translation vector
"""
num_joints = joints_3d.shape[0]
if proj_mat is None:
# focal length
f = np.array([focal_length,focal_length])
# optical center
center = img_size/2.
else:
f = np.array([proj_mat[0,0],proj_mat[1,1]])
center = proj_mat[:2,2]
# transformations
Z = np.reshape(np.tile(joints_3d[:,2],(2,1)).T,-1)
XY = np.reshape(joints_3d[:,0:2],-1)
O = np.tile(center,num_joints)
F = np.tile(f,num_joints)
weight2 = np.reshape(np.tile(np.sqrt(joints_conf),(2,1)).T,-1)
# least squares
Q = np.array([F*np.tile(np.array([1,0]),num_joints), F*np.tile(np.array([0,1]),num_joints), O-np.reshape(joints_2d,-1)]).T
c = (np.reshape(joints_2d,-1)-O)*Z - F*XY
# weighted least squares
W = np.diagflat(weight2)
Q = np.dot(W,Q)
c = np.dot(W,c)
# square matrix
A = np.dot(Q.T,Q)
b = np.dot(Q.T,c)
# solution
trans = np.linalg.solve(A, b)
return trans
def estimate_translation(joints_3d, joints_2d, pts_mnum=4,focal_length=600, proj_mats=None, cam_dists=None,img_size=np.array([512.,512.])):
"""Find camera translation that brings 3D joints joints_3d closest to 2D the corresponding joints_2d.
Input:
joints_3d: (B, K, 3) 3D joint locations
joints: (B, K, 2) 2D joint coordinates
Returns:
(B, 3) camera translation vectors
"""
if torch.is_tensor(joints_3d):
joints_3d = joints_3d.detach().cpu().numpy()
if torch.is_tensor(joints_2d):
joints_2d = joints_2d.detach().cpu().numpy()
if joints_2d.shape[-1]==2:
joints_conf = joints_2d[:, :, -1]>-2.
elif joints_2d.shape[-1]==3:
joints_conf = joints_2d[:, :, -1]>0
joints3d_conf = joints_3d[:, :, -1]!=-2.
trans = np.zeros((joints_3d.shape[0], 3), dtype=np.float)
if proj_mats is None:
proj_mats = [None for _ in range(len(joints_2d))]
if cam_dists is None:
cam_dists = [None for _ in range(len(joints_2d))]
# Find the translation for each example in the batch
for i in range(joints_3d.shape[0]):
S_i = joints_3d[i]
joints_i = joints_2d[i,:,:2]
valid_mask = joints_conf[i]*joints3d_conf[i]
if valid_mask.sum()<pts_mnum:
trans[i] = INVALID_TRANS
continue
if len(img_size.shape)==1:
imgsize = img_size
elif len(img_size.shape)==2:
imgsize = img_size[i]
else:
raise NotImplementedError
try:
trans[i] = estimate_translation_cv2(S_i[valid_mask], joints_i[valid_mask],
focal_length=focal_length, img_size=imgsize, proj_mat=proj_mats[i], cam_dist=cam_dists[i])
except:
trans[i] = estimate_translation_np(S_i[valid_mask], joints_i[valid_mask], valid_mask[valid_mask].astype(np.float32),
focal_length=focal_length, img_size=imgsize, proj_mat=proj_mats[i])
return torch.from_numpy(trans).float()
#-----------------------------------------------------------------------------------------#
# Body joints definition
#-----------------------------------------------------------------------------------------#
def joint_mapping(source_format, target_format):
mapping = np.ones(len(target_format),dtype=np.int)*-1
for joint_name in target_format:
if joint_name in source_format:
mapping[target_format[joint_name]] = source_format[joint_name]
return np.array(mapping)
SMPL_24 = {
'Pelvis_SMPL':0, 'L_Hip_SMPL':1, 'R_Hip_SMPL':2, 'Spine_SMPL': 3, 'L_Knee':4, 'R_Knee':5, 'Thorax_SMPL': 6, 'L_Ankle':7, 'R_Ankle':8,'Thorax_up_SMPL':9, \
'L_Toe_SMPL':10, 'R_Toe_SMPL':11, 'Neck': 12, 'L_Collar':13, 'R_Collar':14, 'Jaw':15, 'L_Shoulder':16, 'R_Shoulder':17,\
'L_Elbow':18, 'R_Elbow':19, 'L_Wrist': 20, 'R_Wrist': 21, 'L_Hand':22, 'R_Hand':23
}
SMPL_EXTRA_30 = {
'Nose':24, 'R_Eye':25, 'L_Eye':26, 'R_Ear': 27, 'L_Ear':28, \
'L_BigToe':29, 'L_SmallToe': 30, 'L_Heel':31, 'R_BigToe':32,'R_SmallToe':33, 'R_Heel':34, \
'L_Hand_thumb':35, 'L_Hand_index': 36, 'L_Hand_middle':37, 'L_Hand_ring':38, 'L_Hand_pinky':39, \
'R_Hand_thumb':40, 'R_Hand_index':41,'R_Hand_middle':42, 'R_Hand_ring':43, 'R_Hand_pinky': 44, \
'R_Hip': 45, 'L_Hip':46, 'Neck_LSP':47, 'Head_top':48, 'Pelvis':49, 'Thorax_MPII':50, \
'Spine_H36M':51, 'Jaw_H36M':52, 'Head':53
}
SMPL_ALL_54 = {**SMPL_24, **SMPL_EXTRA_30}
#-----------------------------------------------------------------------------------------#
# 3D vector to 6D rotation representation conversion utils
#-----------------------------------------------------------------------------------------#
def rot6D_to_angular(rot6D):
batch_size = rot6D.shape[0]
pred_rotmat = rot6d_to_rotmat(rot6D).view(batch_size, -1, 3, 3)
pose = rotation_matrix_to_angle_axis(pred_rotmat.reshape(-1, 3, 3)).reshape(batch_size, -1)
return pose
def rot6d_to_rotmat(x):
x = x.view(-1,3,2)
# Normalize the first vector
b1 = F.normalize(x[:, :, 0], dim=1, eps=1e-6)
dot_prod = torch.sum(b1 * x[:, :, 1], dim=1, keepdim=True)
# Compute the second vector by finding the orthogonal complement to it
b2 = F.normalize(x[:, :, 1] - dot_prod * b1, dim=-1, eps=1e-6)
# Finish building the basis by taking the cross product
b3 = torch.cross(b1, b2, dim=1)
rot_mats = torch.stack([b1, b2, b3], dim=-1)
return rot_mats
def batch_rodrigues(axisang):
# This function is borrowed from https://github.com/MandyMo/pytorch_HMR/blob/master/src/util.py#L37
# axisang N x 3
axisang_norm = torch.norm(axisang + 1e-8, p=2, dim=1)
angle = torch.unsqueeze(axisang_norm, -1)
axisang_normalized = torch.div(axisang, angle)
angle = angle * 0.5
v_cos = torch.cos(angle)
v_sin = torch.sin(angle)
quat = torch.cat([v_cos, v_sin * axisang_normalized], dim=1)
rot_mat = quat2mat(quat)
rot_mat = rot_mat.view(rot_mat.shape[0], 9)
return rot_mat
def quat2mat(quat):
"""
This function is borrowed from https://github.com/MandyMo/pytorch_HMR/blob/master/src/util.py#L50
Convert quaternion coefficients to rotation matrix.
Args:
quat: size = [batch_size, 4] 4 <===>(w, x, y, z)
Returns:
Rotation matrix corresponding to the quaternion -- size = [batch_size, 3, 3]
"""
norm_quat = quat
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]
batch_size = 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).view(batch_size, 3, 3)
return rotMat
def rotation_matrix_to_angle_axis(rotation_matrix):
"""
Convert 3x4 rotation matrix to Rodrigues vector
Args:
rotation_matrix (Tensor): rotation matrix.
Returns:
Tensor: Rodrigues vector transformation.
Shape:
- Input: :math:`(N, 3, 3)`
- Output: :math:`(N, 3)`
Example:
>>> input = torch.rand(2, 3, 3)
>>> output = tgm.rotation_matrix_to_angle_axis(input) # Nx3
"""
quaternion = rotation_matrix_to_quaternion(rotation_matrix)
aa = quaternion_to_angle_axis(quaternion)
aa[torch.isnan(aa)] = 0.0
return aa
def quaternion_to_angle_axis(quaternion: torch.Tensor) -> torch.Tensor:
"""
This function is borrowed from https://github.com/kornia/kornia
Convert quaternion vector to angle axis of rotation.
Adapted from ceres C++ library: ceres-solver/include/ceres/rotation.h
Args:
quaternion (torch.Tensor): tensor with quaternions.
Return:
torch.Tensor: tensor with angle axis of rotation.
Shape:
- Input: :math:`(*, 4)` where `*` means, any number of dimensions
- Output: :math:`(*, 3)`
Example:
>>> quaternion = torch.rand(2, 4) # Nx4
>>> angle_axis = tgm.quaternion_to_angle_axis(quaternion) # Nx3
"""
if not torch.is_tensor(quaternion):
raise TypeError("Input type is not a torch.Tensor. Got {}".format(
type(quaternion)))
if not quaternion.shape[-1] == 4:
raise ValueError("Input must be a tensor of shape Nx4 or 4. Got {}"
.format(quaternion.shape))
# unpack input and compute conversion
q1: torch.Tensor = quaternion[..., 1]
q2: torch.Tensor = quaternion[..., 2]
q3: torch.Tensor = quaternion[..., 3]
sin_squared_theta: torch.Tensor = q1 * q1 + q2 * q2 + q3 * q3
sin_theta: torch.Tensor = torch.sqrt(sin_squared_theta)
cos_theta: torch.Tensor = quaternion[..., 0]
two_theta: torch.Tensor = 2.0 * torch.where(
cos_theta < 0.0,
torch.atan2(-sin_theta, -cos_theta),
torch.atan2(sin_theta, cos_theta))
k_pos: torch.Tensor = two_theta / sin_theta
k_neg: torch.Tensor = 2.0 * torch.ones_like(sin_theta)
k: torch.Tensor = torch.where(sin_squared_theta > 0.0, k_pos, k_neg)
angle_axis: torch.Tensor = torch.zeros_like(quaternion)[..., :3]
angle_axis[..., 0] += q1 * k
angle_axis[..., 1] += q2 * k
angle_axis[..., 2] += q3 * k
return angle_axis
def rotation_matrix_to_quaternion(rotation_matrix, eps=1e-6):
"""
This function is borrowed from https://github.com/kornia/kornia
Convert 3x4 rotation matrix to 4d quaternion vector
This algorithm is based on algorithm described in
https://github.com/KieranWynn/pyquaternion/blob/master/pyquaternion/quaternion.py#L201
Args:
rotation_matrix (Tensor): the rotation matrix to convert.
Return:
Tensor: the rotation in quaternion
Shape:
- Input: :math:`(N, 3, 4)`
- Output: :math:`(N, 4)`
Example:
>>> input = torch.rand(4, 3, 4) # Nx3x4
>>> output = tgm.rotation_matrix_to_quaternion(input) # Nx4
"""
if not torch.is_tensor(rotation_matrix):
raise TypeError("Input type is not a torch.Tensor. Got {}".format(
type(rotation_matrix)))
if len(rotation_matrix.shape) > 3:
raise ValueError(
"Input size must be a three dimensional tensor. Got {}".format(
rotation_matrix.shape))
rmat_t = torch.transpose(rotation_matrix, 1, 2)
mask_d2 = rmat_t[:, 2, 2] < eps
mask_d0_d1 = rmat_t[:, 0, 0] > rmat_t[:, 1, 1]
mask_d0_nd1 = rmat_t[:, 0, 0] < -rmat_t[:, 1, 1]
t0 = 1 + rmat_t[:, 0, 0] - rmat_t[:, 1, 1] - rmat_t[:, 2, 2]
q0 = torch.stack([rmat_t[:, 1, 2] - rmat_t[:, 2, 1],
t0, rmat_t[:, 0, 1] + rmat_t[:, 1, 0],
rmat_t[:, 2, 0] + rmat_t[:, 0, 2]], -1)
t0_rep = t0.repeat(4, 1).t()
t1 = 1 - rmat_t[:, 0, 0] + rmat_t[:, 1, 1] - rmat_t[:, 2, 2]
q1 = torch.stack([rmat_t[:, 2, 0] - rmat_t[:, 0, 2],
rmat_t[:, 0, 1] + rmat_t[:, 1, 0],
t1, rmat_t[:, 1, 2] + rmat_t[:, 2, 1]], -1)
t1_rep = t1.repeat(4, 1).t()
t2 = 1 - rmat_t[:, 0, 0] - rmat_t[:, 1, 1] + rmat_t[:, 2, 2]
q2 = torch.stack([rmat_t[:, 0, 1] - rmat_t[:, 1, 0],
rmat_t[:, 2, 0] + rmat_t[:, 0, 2],
rmat_t[:, 1, 2] + rmat_t[:, 2, 1], t2], -1)
t2_rep = t2.repeat(4, 1).t()
t3 = 1 + rmat_t[:, 0, 0] + rmat_t[:, 1, 1] + rmat_t[:, 2, 2]
q3 = torch.stack([t3, rmat_t[:, 1, 2] - rmat_t[:, 2, 1],
rmat_t[:, 2, 0] - rmat_t[:, 0, 2],
rmat_t[:, 0, 1] - rmat_t[:, 1, 0]], -1)
t3_rep = t3.repeat(4, 1).t()
mask_c0 = mask_d2 * mask_d0_d1
mask_c1 = mask_d2 * ~mask_d0_d1
mask_c2 = ~mask_d2 * mask_d0_nd1
mask_c3 = ~mask_d2 * ~mask_d0_nd1
mask_c0 = mask_c0.view(-1, 1).type_as(q0)
mask_c1 = mask_c1.view(-1, 1).type_as(q1)
mask_c2 = mask_c2.view(-1, 1).type_as(q2)
mask_c3 = mask_c3.view(-1, 1).type_as(q3)
q = q0 * mask_c0 + q1 * mask_c1 + q2 * mask_c2 + q3 * mask_c3
q /= torch.sqrt(t0_rep * mask_c0 + t1_rep * mask_c1 + # noqa
t2_rep * mask_c2 + t3_rep * mask_c3) # noqa
q *= 0.5
return q
def transform_rot_representation(rot, input_type='mat',out_type='vec'):
'''
make transformation between different representation of 3D rotation
input_type / out_type (np.array):
'mat': rotation matrix (3*3)
'quat': quaternion (4)
'vec': rotation vector (3)
'euler': Euler degrees in x,y,z (3)
'''
from scipy.spatial.transform import Rotation as R
if input_type=='mat':
r = R.from_matrix(rot)
elif input_type=='quat':
r = R.from_quat(rot)
elif input_type =='vec':
r = R.from_rotvec(rot)
elif input_type =='euler':
if rot.max()<4:
rot = rot*180/np.pi
r = R.from_euler('xyz',rot, degrees=True)
if out_type=='mat':
out = r.as_matrix()
elif out_type=='quat':
out = r.as_quat()
elif out_type =='vec':
out = r.as_rotvec()
elif out_type =='euler':
out = r.as_euler('xyz', degrees=False)
return out
#-----------------------------------------------------------------------------------------#
# utilizes
#-----------------------------------------------------------------------------------------#
def time_cost(name='ROMP'):
def time_counter(func):
# This function shows the execution time of
# the function object passed
def wrap_func(*args, **kwargs):
t1 = time()
result = func(*args, **kwargs)
t2 = time()
cost_time = t2-t1
fps = 1./cost_time
print(f'{name} {func.__name__!r} executed in {cost_time:.4f}s, FPS {fps:.1f}')
return result
return wrap_func
return time_counter
def determine_device(gpu_id):
if gpu_id != -1:
device = torch.device('cuda:{}'.format(gpu_id))
else:
device = torch.device('cpu')
return device
def download_model(remote_url, local_path, name):
try:
os.makedirs(os.path.dirname(local_path),exist_ok=True)
try:
import wget
except:
print('Installing wget to download model data.')
os.system('pip install wget')
import wget
print('Downloading the {} model from {} and put it to {} \n Please download it by youself if this is too slow...'.format(name, remote_url, local_path))
wget.download(remote_url, local_path)
except Exception as error:
print(error)
print('Failure in downloading the {} model, please download it by youself from {}, and put it to {}'.format(name, remote_url, local_path))
def wait_func(mode):
if mode == 'image':
print('Press ESC to exit...')
while 1:
if cv2.waitKey() == 27:
break
elif mode == 'webcam' or mode == 'video':
cv2.waitKey(1)
class ProgressBar(object):
DEFAULT = 'Progress: %(bar)s %(percent)3d%%'
FULL = "%(bar)s %(current)d/%(total)d (%(percent)3d%%) %(remaining)d to go \n"
def __init__(self, total, width=40, fmt=DEFAULT, symbol='-',
output=sys.stderr):
assert len(symbol) == 1
self.total = total
self.width = width
self.symbol = symbol
self.output = output
self.fmt = re.sub(r'(?P<name>%\(.+?\))d',
r'\g<name>%dd' % len(str(total)), fmt)
self.current = 0
def __call__(self):
percent = self.current / float(self.total)
size = int(self.width * percent)
remaining = self.total - self.current
bar = '[' + self.symbol * size + ' ' * (self.width - size) + ']'
args = {
'total': self.total,
'bar': bar,
'current': self.current,
'percent': percent * 100,
'remaining': remaining
}
print('\r' + self.fmt % args, file=self.output, end='')
def done(self):
self.current = self.total
self()
print('', file=self.output)
def progress_bar(it):
progress = ProgressBar(len(it), fmt=ProgressBar.FULL)
for i, item in enumerate(it):
yield item
progress.current += 1
progress()
progress.done()