-
Notifications
You must be signed in to change notification settings - Fork 2
/
pdr_utils.py
202 lines (150 loc) · 7.13 KB
/
pdr_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
import numpy as np
import pandas as pd
import plotly.graph_objects as go
from plotly.subplots import make_subplots
from datetime import datetime, timedelta
import plotlyHelper
from scipy.signal import find_peaks
import itertools
sensor_types = ['Accelerometer', 'Gyroscope', 'Gravity', 'Barometer', 'Orientation']
class Step:
def __init__(self, owner, length, mag, true_size=None):
self.length = length
self.magnitude = mag
self.true_size = true_size
self.approx_size = None
self.owner = owner
self.raw_signal = None
self.Gk = None
if self.true_size is not None:
self.calc_gain()
def calc_gain(self):
self.Gk = self.true_size / (np.cbrt(self.magnitude / self.length))
def flatten_list(list_of_lists):
return list(itertools.chain(*list_of_lists))
# ------------------------------------------------------ Loading utils ---------------------------------------------
def load_session(dirname, remove_ends_seconds=0, biases=False):
sensor_data = {}
fs = {}
for sensor in sensor_types:
try:
filename = f'data/{dirname}/{sensor}.csv'
df = pd.read_csv(filename, sep=',')
df.sort_values(by='time', inplace=True)
if sensor == 'Accelerometer':
if biases is not False:
for axis, value in biases.items():
df[axis] = df[axis] - value[0]
df['l2_norm'] = np.linalg.norm(df[['x', 'y', 'z']].values, axis=1)
df['str_time'] = pd.to_datetime(df['time'], unit='ns')
sensor_data[sensor] = df
fs[sensor] = np.round(np.mean(1e9 / df['time'].diff()))
if remove_ends_seconds:
sensor_data[sensor] = sensor_data[sensor].iloc[
round(remove_ends_seconds * fs[sensor]):-round(remove_ends_seconds * fs[sensor])]
except IOError:
print("No such sensor for this session")
return sensor_data, fs
# ------------------------------------------------------ Plotting utils ---------------------------------------------
def plot_single_sensor(single_sensor_data, type, fig, col, row, fs=0, peaks=None, peakTH=2):
for metric in single_sensor_data.columns[1:-1]:
fig.add_trace(go.Scatter(x=list(single_sensor_data['str_time']),
y=list(single_sensor_data[metric]),
name=f'{type} ({metric})', mode='lines'),
col=col,
row=row)
if peaks is not None:
fig.add_trace(go.Scatter(x=list(single_sensor_data['str_time'][peaks]),
y=list(single_sensor_data['l2_norm'][peaks]), mode='markers',
marker=dict(size=8, color='red', symbol='cross'),
name='Detected Peaks'),
col=col,
row=row)
# fig.add_trace(go.Scatter(x=[single_sensor_data['str_time'].head(1), single_sensor_data['str_time'].tail(1)],
fig.add_trace(
go.Scatter(x=list(single_sensor_data['str_time'].head(1).append(single_sensor_data['str_time'].tail(1))),
y=[peakTH, peakTH],
mode='lines', line=dict(color="RoyalBlue", width=0.5), fillcolor="LightSkyBlue",
name='TH: ' + str(peakTH)),
row=row, col=col)
fig.layout.annotations[row - 1]['text'] = f'Plot of {type} sensor sampling @ {fs} Hz'
fig.update_xaxes(title='time', tickfont_size=6, **plotlyHelper.axisStyle, row=row, col=1)
fig.update_yaxes(title='Signal', tickfont_size=6, **plotlyHelper.axisStyle, row=row, col=1)
def plot_sensors(sensor_data, fs, peaks=None, title='', peakTH=2):
fig = make_subplots(rows=len(sensor_data), cols=1, horizontal_spacing=0.055,
subplot_titles=['d' for _ in range(len(sensor_data))])
row = 1
for sensor, data in sensor_data.items():
plot_single_sensor(data, sensor, fig, 1, row, fs=fs[sensor],
peaks=peaks if sensor == 'Accelerometer' else None,
peakTH=peakTH)
row += 1
for i in fig['layout']['annotations']:
i['font']['size'] = 14
i['xanchor'] = 'left'
fig.update_layout(title_text=f'<b>Mobile phone sensor readings: {title}</b>', showlegend=True,
**plotlyHelper.layoutStyle)
fig.show()
def plot_steps(list_of_steps):
'''
Plot all steps on-top of each other (just to get a sense of pattern_
:param list_of_steps: list with Step objects
:return: plots the steps aligned on top of each other
'''
fig = go.Figure()
for step in list_of_steps:
fig.add_trace(go.Scatter(x=np.arange(len(step.raw_signal)), y=step.raw_signal,
mode='lines', line=dict(color="RoyalBlue", width=0.5), fillcolor="LightSkyBlue"))
fig.show()
# -------------------------------------------------- Processing utils -------------------------------------------------
def calibrate(acc_vec, n_steps, distance, method='kim'):
if method == 'kim':
sk_true = distance / n_steps
mean_acc = np.cbrt(acc_vec),
gk = sk_true / mean_acc[0]
return gk
else:
print('method not supported')
return False
def collect_steps(data, fs, indices, owner, true_size=None, th=0.58):
steps = []
for i, indice in enumerate(indices[:-1]):
indx_start = max(int(indice - 0.5 * th * fs), 0)
indx_end = min(int(indice + 0.5 * th * fs), len(data['l2_norm']))
abs(data['time'][indices[i + 1]] - data['time'][indices[i]])
mag = sum(list(data['l2_norm'][indx_start:indx_end]))
length = abs(indx_end - indx_start)
steps.append(Step(owner=owner, length=length, mag=mag, true_size=true_size))
steps[-1].raw_signal = list(data['l2_norm'][indx_start:indx_end])
return steps
def calc_mean_gain(steps_list):
flat_list = flatten_list(steps_list)
return sum([step.Gk for step in flat_list]) / len(flat_list)
# ---------------------------------------------------- Testing utils --------------------------------------------------
def calc_per_step_errors(steps_list, gk):
steps_list = flatten_list(steps_list)
error = 0
for step in steps_list:
error += (gk * np.cbrt(step.magnitude / step.length) - step.true_size) ** 2
return np.sqrt(error / len(steps_list))
def calc_per_walk_errors(walking_sessions_list, gk, true_distance=20):
error = 0
for walk in walking_sessions_list:
error += ((true_distance-get_length(walk, gk))/true_distance)**2
return np.sqrt(error / len(walking_sessions_list))
def get_length(steps_list, gk):
length = 0
for step in steps_list:
length += gk * np.cbrt(step.magnitude / step.length)
# print('Calculated Length =', length)
return length
def test(acc_vec, n_steps, gk=None, method='kim'):
if method == 'kim':
if gk:
return np.cbrt(acc_vec) * gk
else:
print('please input Gk gain param')
return False
else:
print('method not supported')
return False