This repository contains measurements pertaining to different human activities. The measurements were made in four batches. The first batch consisted of 8 subjects; the second, of 16 subjects; the third, 10 subjects, and the fourth, 20 subjects. The measurements of Batch 1, Batch 2, and Batch 4 last 120 s; the measurements of Batch 3 last ca. 60 s. All the measurements were sampled at 512 Hz. In the first batch, the activities are bending over, sitting, standing, climbing up and down a staircase (together), walking, and running. In the second, third, and fourth, the activities are: sitting, standing, climbing up a staircase, climbing down a staircase (separated), walking, and running. In addition, the fourth batch adds badminton to the set. There are altogether more than 20 channel. The most important are the following: 3 channels for low-noise acceleration (channels 2 to 4), 3 channels for wide-range acceleration (channels 5 to 7), 3 channels for gyroscope (channels 15 to 17), 5 channels for ECG (channels 10 to 14). The measurements were taken using the Shimmer platform (v. 3). Channels 2:4 correspond to a low-noise 3D accelerometer. These channels are highly sensitive but tend to clip when the activities become intense. Trust these channels for less-intense activity modelling. Channels 5:7 correspond to a wide-band 3D accelerometer and have a wider measurement range, but they are not as sensitive as channels 2:4. The EMG channels were inactive in all the experiments; so the measurements are spurious. Channels 11:14 correspond to measurements of a wireless ECG (LARA: measurements between the left arm and the right arm; LLRA: between left leg and right arm; LLLA: between the left-leg and the left arm. This channel is the least corrupted by noise. Vx is a reference channel). Channels 15:17 correspond to a 3D gyroscope.
-
Notifications
You must be signed in to change notification settings - Fork 0
netd-tud/human-activity-Recognition
About
Activity Recognition using IMU and Wireless ECG
Topics
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published