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
| class AlcoholImpairmentDetector:
"""
Smart Eye酒精损伤检测模型
检测指标：
1. 眼动模式：扫视速度下降、凝视时间增加
2. 面部特征：面部肌肉松弛、表情呆滞
3. 头部运动：反应延迟、微动作减少
"""
def __init__(self):
self.gaze_analyzer = GazePatternAnalyzer()
self.expression_analyzer = FacialExpressionAnalyzer()
self.head_motion_analyzer = HeadMotionAnalyzer()
self.fusion_classifier = MultiModalFusion()
def detect(self, frames, gaze_data, head_pose_data):
"""
检测酒精损伤状态
Args:
frames: (T, 3, H, W) 视频序列
gaze_data: (T, 3) 视线方向序列
head_pose_data: (T, 3) 头部姿态序列
Returns:
impairment_prob: 酒精损伤概率
confidence: 置信度
"""
gaze_features = self.gaze_analyzer.extract_patterns(gaze_data)
expression_features = self.expression_analyzer.analyze(frames)
motion_features = self.head_motion_analyzer.analyze(head_pose_data)
features = {
'gaze': gaze_features,
'expression': expression_features,
'motion': motion_features
}
impairment_prob = self.fusion_classifier.predict(features)
return impairment_prob
class GazePatternAnalyzer:
"""
眼动模式分析器
酒精损伤特征：
- 扫视速度下降30-50%
- 凝视时间增加
- 视线切换频率降低
"""
def extract_patterns(self, gaze_sequence):
"""
提取眼动模式特征
Args:
gaze_sequence: (T, 3) 视线方向序列
Returns:
features: dict
"""
saccade_velocity = self._compute_saccade_velocity(gaze_sequence)
fixation_duration = self._compute_fixation_duration(gaze_sequence)
gaze_switch_rate = self._compute_gaze_switch_rate(gaze_sequence)
return {
'saccade_velocity': saccade_velocity,
'fixation_duration': fixation_duration,
'gaze_switch_rate': gaze_switch_rate
}
def _compute_saccade_velocity(self, gaze_seq):
"""计算扫视速度（度/秒）"""
delta = torch.diff(gaze_seq, dim=0)
angular_change = torch.norm(delta, dim=-1)
velocity = torch.mean(angular_change).item()
return velocity
def _compute_fixation_duration(self, gaze_seq):
"""计算平均凝视时间（毫秒）"""
velocity = torch.norm(torch.diff(gaze_seq, dim=0), dim=-1)
fixation_mask = velocity < 0.01
fixation_durations = []
current_duration = 0
for is_fixation in fixation_mask:
if is_fixation:
current_duration += 33
else:
if current_duration > 0:
fixation_durations.append(current_duration)
current_duration = 0
return np.mean(fixation_durations) if fixation_durations else 0
|