Activity recognition dataset - Skoda Mini Checkpoint

Daniel Roggen, Wearable Computing Laboratory, ETH Zurich
droggen@gmail.com
Initial documentation: 24.03.2009
Last updated: 06.08.2015

Description

This dataset contains 10 manipulative gestures performed in a car maintenance scenario.
They are a subset of the 46 activities performed in the factory in one of the quality control checkpoints [Stiefmeier07]

This dataset was initially collected to investigate the use of ensemble classifiers (fusion of multiple classifier decision operating on individual sensor nodes) in activity recognition.
In particular: resilience to faults; performance scaling with number of sensors, power-performance management [Zappi07,Zappi08,Zappi12].

Availability

This dataset can be freely used in publications provided one of the following paper is cited: [Zappi08] or [Zappi12].

Sensors

This dataset contains 10 classes, recorded with a 2x10 USB sensors placed on the left and right upper and lower arm.

Sensor sample rate is approximately 98Hz.
The locations of the sensors on the arms is documented in the figure.
Location of the sensor on the body

Activities

10 activiteis


Files

Dataset original cleaned


right_classall_clean.mat and left_classall_clean.mat: matlab .mat files with original datafor right and left arm sensors

Load into matlab with the load command.

Variables loaded: right_classall_clean and left_classall_clean

Matrix format: one line per sample.


with s=0...29 are the sensor axis (10 3-axis sensors = 10x3 = 30 axis). Sensor node number is mod(s,3).
Calibrated acceleration means acceleration in milli-g units (1000 = earth gravity vector).
Raw acceleration is ADC readout.

label value:

Dataset segmented

dataset_cp_2007_12.mat: segmented dataset

Load with matlab load command.

Variables loaded: dataset_left, dataset_right

Variable format:

dataset_xx{axisnumber}{classnumber}{instancenumber} is a vector with calibrated acceleration data; length of vector is gesture length dependent.

Mapping of axisnumber to sensor number:
left axisnumber 1-3 = sensor 3 x,y,z
left axisnumber 4-6 = sensor 17 x,y,z
left axisnumber 7-9 = sensor 19 x,y,z
left axisnumber 10-12 = sensor 20 x,y,z
left axisnumber 13-15 = sensor 23 x,y,z
left axisnumber 16-18 = sensor 25 x,y,z
left axisnumber 19-21 = sensor 26 x,y,z
left axisnumber 22-24 = sensor 28 x,y,z
left axisnumber 25-27 = sensor 30 x,y,z
left axisnumber 28-30 = sensor 31 x,y,z

right axisnumber 1-3 = sensor 1 x,y,z
right axisnumber 4-6 = sensor 2 x,y,z
right axisnumber 7-9 = sensor 14 x,y,z
right axisnumber 10-12 = sensor 16 x,y,z
right axisnumber 13-15 = sensor 18 x,y,z
right axisnumber 16-18 = sensor 21 x,y,z
right axisnumber 19-21 = sensor 22 x,y,z
right axisnumber 22-24 = sensor 24 x,y,z
right axisnumber 25-27 = sensor 27 x,y,z
right axisnumber 28-30 = sensor 29 x,y,z

References