- #Simple android app to get accelerometer data manual
- #Simple android app to get accelerometer data free
The noisy measurements and the beginning and ending are the result of starting and stopping the logging while the phone is in the hand. The annotations are added manually to aid the interpretation of the data for the reader. Without any knowledge about the performed activities, a (slightly trained) human eye can recognize different activities, or at least transitions between them. ResultsĪn example of a gathered plot can be seen here. transform_logs_to_directory.rb logs/set-stand-walk-stand-sitĪn installed Ruby and Octave environment is required.
These steps can be executed by using the command line script: This script will create a graph for each measurement and one file with a few graphs accumulated. The next step is to plot the data and save the graphs is png format.įor this the directory with the transformed csv files is passed on to a Octave script. This is done by a Ruby script, which scans a directory for Sensor Logger created files. The first step of this process is to strip the csv files from comments and prepare them for an easy Octave load call. When the data is gathered on the computer, it needs to be transformed and plotted.įor this we use a combination of Ruby and Octave scripts. by USB or bluetooth.īecause our wish is fast data-gathering, DropSpace is used to mirror the SD-card directory to our computer, using the Dropbox system. These files can be transmitted to a PC, e.g. The Sensor Logger app logs the sensor data to eight separate csv files on the SD-card of the smartphone. Its benefits are the adjustable sensor rate, the number of (virtual) sensors (raw accelerometer, linear acceleration, orientation, rotation, etc) and the easy csv format in which data is logged.Ī drawback is the lack of gps recording, but due the nature of your recordings (small in- or outdoor activities), this is not a big problem for now. This implementation simplicity is reflected in the number of apps available performing the task of recording sensory data.Īfter a quick review of available apps in the Google Play store, our weapon of choice became Sensor Logger. Android AppsĪlthough the implementation of an Android app to gather the accelerometer data is not that complicated, it is easier to choose an existing solution. This way we are able to get a feel of the data, before formalizing the idea. The requirements are to quickly perform an activity (without an extensive environment setup) and visualise the data for human inspection.
#Simple android app to get accelerometer data manual
To overcome this problem of unknown factors, we want to gather our own data.īecause we will do small runs of a few consecutive performed activities, such as sitting, walking and taking the stairs, manual annotation would suffice and there is no need for precise labeling of each measurement.
#Simple android app to get accelerometer data free
In our first attempt we used free and open available datasets with labeled activity data the dataset of Human Activity Recognition Using Smarthphones from the UCI Machine Learning Repository and the WISDM dataset.Īlthough it is a luxury to have labeled data, any uncertainty about performed activities and conditions is still a drawback. To try out this idea, one can work with artificial data or, as we liked to do, with real world data.
The hypotheses is that by segmenting the gathered sensory data before applying classification, as is often done directly, a richer and better understanding of the data can be obtained. We are working on a system that can be seen as a part of a pipeline process of recognizing human activities using smartphone data. Using a simple setup this is made possible, for everybody with an Android smartphone.
Every major application or breakthrough starts with just a simple idea.īut an idea that has developed in the mind of one needs to tested to the real world,Īnd preferably in an easy and fast manner.ĭuring a research project on recognizing human activities using accelerometer data, the need for quick visualizations of collected data emerged.
0 kommentar(er)
