Dataset Open Access

Occupancy Sensing and Activity Recognition with Cameras and Wireless Sensors

Zhao, Yang

This dataset contains human activity data from a wireless sensing system, which includes a Doppler motion sensor and a wireless network. The Doppler sensor is a low-cost dual Doppler sensor modified from a commercial-off-the-shelf range-controlled radar, which operates at 5.8 GHz with two directional antennas. The wireless network uses four IEEE 802.15.4 radio nodes (CC2531 from TI) to create a mesh network to measure the RSS between each pair of radio nodes operating on the 16 frequency channels at 2.4 GHz.

For the activity experiment, we recruited human subjects to perform 42 trials of four activities  (each one with two minutes duration): (1)  walking in a room (10 trials), (2) sitting in a chair (10 trials), (3) lying on a bed (12 trials), and (4) body turning on a bed (10 trials). For the walking activity, the human subjects walk along different paths at different locations in the room. For the lying on bed activity, we ask human subjects to breathe normally on bed with three orientations facing upwards, right and left. Finally, for the turning on bed case, human subjects turn their bodies from one side to the other on bed with random time intervals. We also recorded two-minute data of the empty room case before and after each human subject trial. Note that each data file name has its corresponding activity in it, so it is pretty self-explanatory. 

Files (7.0 MB)
Name Size
71.6 kB Download
6.9 MB Download
All versions This version
Views 8282
Downloads 2222
Data volume 90.7 MB90.7 MB
Unique views 7373
Unique downloads 1313


Cite as