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
| import cv2
import numpy as np
from typing import Tuple, List, Dict
class OOPDetector:
"""
基于视觉的Out-of-Position检测
"""
def __init__(self):
self.pose_estimator = PoseEstimator()
self.normal_pose_threshold = {
'forward_lean': 15,
'lateral_lean': 10,
'head_distance': 300,
}
def detect_oop(self, frame: np.ndarray) -> Dict:
"""
检测OOP状态
Args:
frame: 输入图像
Returns:
oop_status: OOP检测结果
"""
keypoints = self.pose_estimator.detect(frame)
if keypoints is None:
return {'is_oop': False, 'reason': 'no_person'}
forward_lean = self._calculate_forward_lean(keypoints)
lateral_lean = self._calculate_lateral_lean(keypoints)
head_position = self._get_head_position(keypoints)
oop_flags = []
if forward_lean > self.normal_pose_threshold['forward_lean']:
oop_flags.append(f'forward_lean: {forward_lean:.1f}°')
if abs(lateral_lean) > self.normal_pose_threshold['lateral_lean']:
oop_flags.append(f'lateral_lean: {lateral_lean:.1f}°')
head_distance = self._estimate_head_to_dashboard(head_position, frame.shape)
if head_distance < self.normal_pose_threshold['head_distance']:
oop_flags.append(f'too_close: {head_distance:.0f}mm')
is_oop = len(oop_flags) > 0
return {
'is_oop': is_oop,
'flags': oop_flags,
'forward_lean': forward_lean,
'lateral_lean': lateral_lean,
'head_distance': head_distance,
}
def _calculate_forward_lean(self, keypoints: Dict) -> float:
"""
计算前倾角度
Args:
keypoints: 关键点字典
Returns:
forward_lean: 前倾角度（度）
"""
shoulder = keypoints.get('left_shoulder')
hip = keypoints.get('left_hip')
ear = keypoints.get('left_ear')
if None in [shoulder, hip, ear]:
return 0.0
shoulder_hip = np.array(shoulder) - np.array(hip)
vertical = np.array([0, 1])
cos_angle = np.dot(shoulder_hip, vertical) / np.linalg.norm(shoulder_hip)
angle = np.arccos(np.clip(cos_angle, -1, 1))
return np.degrees(angle)
def _calculate_lateral_lean(self, keypoints: Dict) -> float:
"""
计算侧倾角度
"""
left_shoulder = keypoints.get('left_shoulder')
right_shoulder = keypoints.get('right_shoulder')
if None in [left_shoulder, right_shoulder]:
return 0.0
shoulder_vector = np.array(right_shoulder) - np.array(left_shoulder)
horizontal = np.array([1, 0])
cos_angle = np.dot(shoulder_vector, horizontal) / np.linalg.norm(shoulder_vector)
angle = np.arccos(np.clip(cos_angle, -1, 1))
if shoulder_vector[1] > 0:
return np.degrees(angle)
else:
return -np.degrees(angle)
def _get_head_position(self, keypoints: Dict) -> Tuple[float, float]:
"""获取头部位置"""
nose = keypoints.get('nose')
if nose:
return nose
return (0, 0)
def _estimate_head_to_dashboard(self, head_pos: Tuple, frame_shape: Tuple) -> float:
"""
估计头部到仪表盘距离
简化方法：基于头部在图像中的垂直位置
"""
height, width = frame_shape[:2]
head_y = head_pos[1]
dashboard_y = height * 0.8
pixel_distance = dashboard_y - head_y
mm_per_pixel = 5
return pixel_distance * mm_per_pixel
class PoseEstimator:
"""姿态估计器（简化版）"""
def __init__(self):
self.mp_pose = None
def detect(self, frame: np.ndarray) -> Dict:
"""
检测人体关键点
Returns:
keypoints: 关键点字典
"""
return {
'nose': (320, 100),
'left_ear': (300, 90),
'right_ear': (340, 90),
'left_shoulder': (280, 200),
'right_shoulder': (360, 200),
'left_hip': (290, 350),
'right_hip': (350, 350),
}
|