-
Notifications
You must be signed in to change notification settings - Fork 25
/
pose.py
203 lines (180 loc) · 6.93 KB
/
pose.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
# -*- coding: utf-8 -*-
# ---------------------
from typing import *
import cv2
import numpy as np
from joint import Joint
class Pose(list):
"""
a Pose is a list of Joint(s) belonging to the same person.
"""
LIMBS = [
(0, 1), # head_top -> head_center
(1, 2), # head_center -> neck
(2, 3), # neck -> right_clavicle
(3, 4), # right_clavicle -> right_shoulder
(4, 5), # right_shoulder -> right_elbow
(5, 6), # right_elbow -> right_wrist
(2, 7), # neck -> left_clavicle
(7, 8), # left_clavicle -> left_shoulder
(8, 9), # left_shoulder -> left_elbow
(9, 10), # left_elbow -> left_wrist
(2, 11), # neck -> spine0
(11, 12), # spine0 -> spine1
(12, 13), # spine1 -> spine2
(13, 14), # spine2 -> spine3
(14, 15), # spine3 -> spine4
(15, 16), # spine4 -> right_hip
(16, 17), # right_hip -> right_knee
(17, 18), # right_knee -> right_ankle
(15, 19), # spine4 -> left_hip
(19, 20), # left_hip -> left_knee
(20, 21) # left_knee -> left_ankle
]
SKELETON = [[l[0] + 1, l[1] + 1] for l in LIMBS]
def __init__(self, joints, person_id=None):
# type: (List[Joint]) -> None
super().__init__(joints)
self.person_id = int(person_id)
@property
def invisible(self):
# type: () -> bool
"""
:return: True if all the joints of the pose are occluded, False otherwise
"""
for j in self:
if not j.occ:
return False
return True
@property
def bbox_2d(self):
# type: () -> List[int]
"""
:return: bounding box around the pose in format [x_min, y_min, width, height]
- x_min = x of the top left corner of the bounding box
- y_min = y of the top left corner of the bounding box
"""
x_min = int(np.min([j.x2d for j in self]))
y_min = int(np.min([j.y2d for j in self]))
x_max = int(np.max([j.x2d for j in self]))
y_max = int(np.max([j.y2d for j in self]))
width = x_max - x_min
height = y_max - y_min
return [x_min, y_min, width, height]
@property
def bbox_head_2d(self):
# type: () -> List[int]
"""
The bbox_head is computed by tacking a square centered around the head with
sides of length equal to two times the distance between the head_top and head_center.
WARNING: it may be wrong !
:return: bounding box around the head in format [x_min, y_min, width, height]
- x_min = x of the top left corner of the head bounding box
- y_min = y of the top left corner of the head bounding box
"""
bbox_length = 2 * np.sqrt((self[0].x2d - self[1].x2d) ** 2 + (self[0].y2d - self[1].y2d) ** 2)
x_min = int(self[1].x2d - bbox_length / 2)
y_min = int(self[1].y2d - bbox_length / 2)
width = int(bbox_length)
height = int(bbox_length)
return [x_min, y_min, width, height]
@property
def bbox_2d_padded(self, h_inc_perc=0.15, w_inc_perc=0.1):
x_min = int(np.min([j.x2d for j in self]))
y_min = int(np.min([j.y2d for j in self]))
x_max = int(np.max([j.x2d for j in self]))
y_max = int(np.max([j.y2d for j in self]))
width = x_max - x_min
height = y_max - y_min
inc_h = (height * h_inc_perc) / 2
inc_w = (width * w_inc_perc) / 2
x_min = x_min - inc_w
x_max = x_max + inc_w
y_min = y_min - inc_h
y_max = y_max + inc_h
width = x_max - x_min
height = y_max - y_min
return [x_min, y_min, width, height]
@property
def coco_annotation(self):
# type: () -> Dict
"""
:return: COCO annotation dictionary of the pose
==========================================================
NOTE#1: in COCO, each keypoint is represented by its (x,y)
2D location and a visibility flag `v` defined as:
- `v=0` ==> not labeled (in which case x=y=0)
- `v=1` ==> labeled but not visible
- `v=2` ==> labeled and visible
==========================================================
NOTE#2: in COCO, a keypoint is considered visible if it
falls inside the object segment. In JTA there are no
object segments and every keypoint is labelled, so we
v=2 for each keypoint.
==========================================================
"""
keypoints = []
for j in self:
keypoints += [j.x2d, j.y2d, 2]
annotation = {
'keypoints': keypoints,
'num_keypoints': len(self),
'bbox': self.bbox_2d
}
return annotation
@property
def posetrack_annotation(self):
# type: () -> Dict
"""
:return: PoseTrack annotation dictionary of the pose
==========================================================
NOTE#1: in PoseTrack, each keypoint is represented by its (x,y)
2D location and a visibility flag `v` defined as:
- `v=0` ==> not visible
- `v=1` ==> visible
==========================================================
NOTE#2: in PoseTrack, keypoints that are not visible are not labeled and
set to (x=y=v=0). In JTA every keypoint is labelled, so keypoints with
v = 0 are labelled with their true (x,y) location.
==========================================================
NOTE#3: in PoseTrack, the bbox_head attribute has been human annotated.
The bbox_head is computed by tacking a square centered around the head with
sides of length equal to two times the distance between the head_top and
head_center.
WARNING: it may be wrong !
==========================================================
"""
keypoints = []
for j in self:
if j.occ:
keypoints += [j.x2d, j.y2d, 0]
else:
keypoints += [j.x2d, j.y2d, 1]
annotation = {
'bbox_head': self.bbox_head_2d,
'keypoints': keypoints,
'bbox': self.bbox_2d,
'track_id': self.person_id,
}
return annotation
def draw(self, image, color):
# type: (np.ndarray, List[int]) -> np.ndarray
"""
:param image: image on which to draw the pose
:param color: color of the limbs make up the pose
:return: image with the pose
"""
# draw limb(s) segments
for (j_id_a, j_id_b) in Pose.LIMBS:
joint_a = self[j_id_a] # type: Joint
joint_b = self[j_id_b] # type: Joint
t = 1 if joint_a.cam_distance > 25 else 2
if joint_a.is_on_screen and joint_b.is_on_screen:
cv2.line(image, joint_a.pos2d, joint_b.pos2d, color=color, thickness=t)
# draw joint(s) circles
for joint in self:
image = joint.draw(image)
return image
def __iter__(self):
# type: () -> Iterator[Joint]
return super().__iter__()