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
| import numpy as np
from scipy.stats import entropy
def extract_gaze_features(gaze_x, gaze_y, fixations, saccades, pupil_diameter):
"""
从IDD数据集提取损伤检测关键特征
Args:
gaze_x: (N,) 水平视线位置 0-1
gaze_y: (N,) 垂直视线位置 0-1
fixations: list of (start, end, duration)
saccades: list of (start, end, amplitude)
pupil_diameter: (N,) 瞳孔直径 mm
Returns:
features: dict, 用于损伤分类的特征向量
"""
features = {}
hist_2d, _, _ = np.histogram2d(gaze_x, gaze_y, bins=20)
hist_2d_norm = hist_2d / hist_2d.sum()
features['gaze_entropy'] = entropy(hist_2d_norm.flatten())
fix_durations = [f[2] for f in fixations]
features['fixation_count'] = len(fixations)
features['mean_fixation_duration'] = np.mean(fix_durations)
features['fixation_rate'] = len(fixations) / (len(gaze_x) / 60)
saccade_amplitudes = [s[2] for s in saccades]
features['saccade_count'] = len(saccades)
features['mean_saccade_amplitude'] = np.mean(saccade_amplitudes) if saccade_amplitudes else 0
features['pupil_mean'] = np.mean(pupil_diameter)
features['pupil_std'] = np.std(pupil_diameter)
features['pupil_trend'] = np.polyfit(range(len(pupil_diameter)),
pupil_diameter, 1)[0]
features['gaze_x_std'] = np.std(gaze_x)
features['gaze_y_std'] = np.std(gaze_y)
features['gaze_spread'] = np.sqrt(
np.var(gaze_x) + np.var(gaze_y)
)
return features
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import cross_val_score
def train_impairment_classifier(features_dict, labels):
"""
使用IDD提取的特征训练损伤分类器
Args:
features_dict: list of feature dicts
labels: 0=正常, 1=认知分心, 2=酒精损伤, 3=复合
Returns:
clf: 训练好的分类器
"""
feature_names = list(features_dict[0].keys())
X = np.array([[f[k] for k in feature_names] for f in features_dict])
y = np.array(labels)
clf = RandomForestClassifier(n_estimators=100, max_depth=10, random_state=42)
scores = cross_val_score(clf, X, y, cv=10, scoring='f1_macro')
print(f"损伤分类器性能:")
print(f" 特征数: {len(feature_names)}")
print(f" 10折交叉验证 F1: {scores.mean():.3f} ± {scores.std():.3f}")
clf.fit(X, y)
importances = sorted(zip(feature_names, clf.feature_importances_),
key=lambda x: -x[1])
print(f"\n特征重要性排序:")
for name, imp in importances:
print(f" {name}: {imp:.4f}")
return clf
if __name__ == "__main__":
np.random.seed(42)
n_samples = 60 * 60
normal_gaze_x = np.random.normal(0.5, 0.15, n_samples)
normal_gaze_y = np.random.normal(0.3, 0.12, n_samples)
normal_pupil = np.random.normal(4.0, 0.3, n_samples)
cog_gaze_x = np.random.normal(0.5, 0.08, n_samples)
cog_gaze_y = np.random.normal(0.25, 0.06, n_samples)
cog_pupil = np.random.normal(4.2, 0.4, n_samples)
alc_gaze_x = np.random.normal(0.5, 0.05, n_samples)
alc_gaze_y = np.random.normal(0.2, 0.04, n_samples)
alc_pupil = np.random.normal(3.8, 0.2, n_samples)
print("正常驾驶特征:")
normal_feat = extract_gaze_features(
normal_gaze_x, normal_gaze_y,
[(0, 60, 0.3)] * 80, [(60, 65, 5)] * 40,
normal_pupil
)
for k, v in normal_feat.items():
print(f" {k}: {v:.4f}")
print("\n认知分心特征:")
cog_feat = extract_gaze_features(
cog_gaze_x, cog_gaze_y,
[(0, 60, 0.5)] * 50, [(60, 65, 3)] * 25,
cog_pupil
)
for k, v in cog_feat.items():
print(f" {k}: {v:.4f}")
print("\n酒精损伤特征:")
alc_feat = extract_gaze_features(
alc_gaze_x, alc_gaze_y,
[(0, 60, 0.7)] * 30, [(60, 65, 2)] * 15,
alc_pupil
)
for k, v in alc_feat.items():
print(f" {k}: {v:.4f}")
print("\n" + "=" * 50)
print("隧道视觉量化 (gaze_spread):")
print(f" 正常: {normal_feat['gaze_spread']:.4f}")
print(f" 认知分心: {cog_feat['gaze_spread']:.4f} (×{cog_feat['gaze_spread']/normal_feat['gaze_spread']:.2f})")
print(f" 酒精损伤: {alc_feat['gaze_spread']:.4f} (×{alc_feat['gaze_spread']/normal_feat['gaze_spread']:.2f})")
|
