Conditions of use
The following videos and data are for research purposes only. No commercial uses authorized without written consent. Any researchers wishing to use these videos for publications should contact Alex Edgcomb (aedgcomb@cs.ucr.edu) and Frank Vahid (vahid@cs.ucr.edu) for citation information.
This data set contains:
- 87 1-minute recordings of a 27 year old male in an at-home environment. The 87 recordings are further categorized:
- 18 confounding scenarios (e.g., looking under couch and fall next to upright vacuum that was just used)
- 5 including a fall
- 13 not including a fall
- 69 non-confounding scenarios (e.g., watching TV and sweeping floor)
- 14 scenarios used to train algorithm
- 9 including a fall
- 5 not including a fall
- 55 scenarios used to test algorithm
- 26 including a fall
- 29 not including a fall
- 4 types of data are included for each recording:
- Video recording at 15FPS with 352x240 resolution
- 3D head height tracking, manually collected at 15 samples/sec
- 2D head position tracking, manually collected at 15 samples/sec
- Minimum bounding rectangle (MBR) around person, automatically collected via computer vision algorithm at 15 samples/sec