Python package for analyzing sensor-collected human motion data

Last update: Nov 05, 2022

Overview

Installation | Requirements | Usage | Contribution | Getting Help

Sensor Motion

https://readthedocs.org/projects/sensormotion/badge/?version=latest

https://badges.gitter.im/gitterHQ/gitter.png

Python package for analyzing sensor-collected human motion data (e.g. physical activity levels, gait dynamics).

Dedicated accelerometer devices, such as those made by Actigraph, usually bundle software for the analysis of the sensor data. In my work I often collect sensor data from smartphones and have not been able to find any comparable analysis software.

This Python package allows the user to extract human motion data, such as gait/walking dynamics, directly from accelerometer signals. Additionally, the package allows for the calculation of physical activity (PA) or moderate-to-vigorous physical activity (MVPA) counts, similar to activity count data offered by companies like Actigraph.

Installation

You can install this package using pip:

pip install sensormotion

Requirements

This package has the following dependencies, most of which are just Python packages:

Python 3.x
- The easiest way to install Python is using the Anaconda distribution, as it also includes the other dependencies listed below
- Python 2.x has not been tested, so backwards compatibility is not guaranteed
numpy
- Included with Anaconda. Otherwise, install using pip (pip install numpy)
scipy
- Included with Anaconda. Otherwise, install using pip (pip install scipy)
matplotlib
- Included with Anaconda. Otherwise, install using pip (pip install matplotlib)

Usage

Here is brief example of extracting step-based metrics from raw vertical acceleration data:

Import the package:

import sensormotion as sm

If you have a vertical acceleration signal x, and its corresponding time signal t, we can begin by filtering the signal using a low-pass filter:

b, a = sm.signal.build_filter(frequency=10,
                              sample_rate=100,
                              filter_type='low',
                              filter_order=4)

x_filtered = sm.signal.filter_signal(b, a, signal=x)

Next, we can detect the peaks (or valleys) in the filtered signal, which gives us the time and value of each detection. Optionally, we can include a plot of the signal and detected peaks/valleys:

peak_times, peak_values = sm.peak.find_peaks(time=t, signal=x_filtered,
                                             peak_type='valley',
                                             min_val=0.6, min_dist=30,
                                             plot=True)

From the detected peaks, we can then calculate step metrics like cadence and step time:

cadence = sm.gait.cadence(time=t, peak_times=peak_times, time_units='ms')
step_mean, step_sd, step_cov = sm.gait.step_time(peak_times=peak_times)

Physical activity counts and intensities can also be calculated from the acceleration data:

x_counts = sm.pa.convert_counts(x, time, integrate='simpson')
y_counts = sm.pa.convert_counts(y, time, integrate='simpson')
z_counts = sm.pa.convert_counts(z, time, integrate='simpson')
vm = sm.signal.vector_magnitude(x_counts, y_counts, z_counts)
categories, time_spent = sm.pa.cut_points(vm, set_name='butte_preschoolers', n_axis=3)

For a more in-depth tutorial, and more workflow examples, please take a look at the tutorial.

I would also recommend looking over the documentation to see other functionalities of the package.

Contribution

I work on this package in my spare time, on an "as needed" basis for my research projects. However, pull requests for bug fixes and new features are always welcome!

Please see the develop branch for the development version of the package, and check out the issues page for bug reports and feature requests.

Getting Help

You can find the full documentation for the package here

Python's built-in help function will show documentation for any module or function: help(sm.gait.step_time)

You're encouraged to post questions, bug reports, or feature requests as an issue

Alternatively, ask questions on Gitter

Comments

Question

I am using sensormotion.py package for finding peaks for one of my applications. I want to know how normalized min_value (0-1) in peak.find_peaks is related to minimum detectable peak value.

opened by vivekmahadev 2
I need help using this library!

Hi

I'm very interested in using this library in my project. I have a test of 2min walking at 100Hz and I collect the data from accelerometer, gyro and magnetometer of an Iphone 6.

I'm trying to use the library with my data but I could understand some things. For example this function sm.peak.find_peaks(ac_lags, ac, peak_type='peak', min_val= 0.6, min_dist=32, plot=True). What are the suitable values of min_val and min_dist parameters? Are they problem dependent? I have tried with many values and the step estimation is not correct.

Please, could you help me?

Best regards

opened by ogreyesp 1
sm.gait.step_regularity IndexError

step_reg, stride_reg = sm.gait.step_regularity(ac_peak_values) File ".../python3.6/site-packages/sensormotion-1.1.0-py3.6.egg/sensormotion/gait.py", line 128, in step_regularity ac_d2 = peaks_half[2] # second dominant period i.e. a stride (left-left) sm.gait.step_regularity IndexError: index 2 is out of bounds for axis 0 with size 2

opened by jiakang 1
Example: Importing from live cvs file?

Do you have any examples where a 1.) step is to open a demo cvs file, and parse the necessary data to be able to apply your package to the data?

I'm needing help starting from square one.

Android-Go01_MetaWear-01_2022-03-26T11.14.56.138_F5EC5C968CB3_Accelerometer_1.3.6.csv

opened by RandoSY 1
Question about step regularity

Hey, I'm using your package right now to generate features for a dataset. I have looked at the paper by Moe Nilssen et al. and tried to follow the steps for calculating step and stride regularity. However, I wonder why you still do the following calculation at the end:

step_reg = ac_d1 / ac_lag0 stride_reg = ac_d2 / ac_lag0

Can you help me with this?

opened by vanessabin 1

Releases(1.1.4)

1.1.4(Nov 12, 2018)
Code formatting

Source code(tar.gz)
Source code(zip)
1.1.3(Oct 21, 2018)

Added warnings for short signal lengths
Source code(tar.gz)
Source code(zip)
1.1.2(Oct 21, 2018)

Sphinx doc update
Source code(tar.gz)
Source code(zip)
1.1.1(Oct 21, 2018)

Fix pip install from pypi
Source code(tar.gz)
Source code(zip)
1.1.0(Apr 21, 2018)

Added conversion of accelerometer data to physical activity counts and intensities
Source code(tar.gz)
Source code(zip)
1.0.3(Apr 19, 2018)

Drastically improved documentation
Source code(tar.gz)
Source code(zip)
1.0.2(Apr 19, 2018)

Source code(tar.gz)
Source code(zip)
1.0.1(Apr 19, 2018)

Source code(tar.gz)
Source code(zip)
1.0.0(Apr 19, 2018)

Added gait analysis and signal processing
Source code(tar.gz)
Source code(zip)

Owner

Simon Ho

Data Science | Machine Learning | Statistics | Gaming

GitHub Repository

In this tutorial, raster models of soil depth and soil water holding capacity for the United States will be sampled at random geographic coordinates within the state of Colorado.

Raster_Sampling_Demo (Resulting graph of this demo) Background Sampling values of a raster at specific geographic coordinates can be done with a numbe

2 Dec 13, 2022

Python package for analyzing sensor-collected human motion data

Related tags

Overview

Sensor Motion

Installation

Requirements

Usage

Contribution

Getting Help

Comments

Question

I need help using this library!

sm.gait.step_regularity IndexError

Example: Importing from live cvs file?

Question about step regularity

Releases(1.1.4)

1.1.4(Nov 12, 2018)

1.1.3(Oct 21, 2018)

1.1.2(Oct 21, 2018)

1.1.1(Oct 21, 2018)

1.1.0(Apr 21, 2018)

1.0.3(Apr 19, 2018)

1.0.2(Apr 19, 2018)

1.0.1(Apr 19, 2018)

1.0.0(Apr 19, 2018)

Owner

Simon Ho

In this tutorial, raster models of soil depth and soil water holding capacity for the United States will be sampled at random geographic coordinates within the state of Colorado.

We're Team Arson and we're using the power of predictive modeling to combat wildfires.

Using Python to derive insights on particular Pokemon, Types, Generations, and Stats

Single machine, multiple cards training; mix-precision training; DALI data loader.

The lastest all in one bombing tool coded in python uses tbomb api

This repository contains some analysis of possible nerdle answers

fds is a tool for Data Scientists made by DAGsHub to version control data and code at once.

A Python 3 library making time series data mining tasks, utilizing matrix profile algorithms

Spectral Analysis in Python

Bamboolib - a GUI for pandas DataFrames

A crude Hy handle on Pandas library

Geospatial data-science analysis on reasons behind delay in Grab ride-share services

The micro-framework to create dataframes from functions.

Pypeln is a simple yet powerful Python library for creating concurrent data pipelines.

An Indexer that works out-of-the-box when you have less than 100K stored Documents

A probabilistic programming language in TensorFlow. Deep generative models, variational inference.

Data and code accompanying the paper Politics and Virality in the Time of Twitter

Python Library for learning (Structure and Parameter) and inference (Statistical and Causal) in Bayesian Networks.

Data science/Analysis Health Care Portfolio

Flenser is a simple, minimal, automated exploratory data analysis tool.