Intela shows progress in people analytics from video
City planners, space designers, retailers and building owners all want to know how people interact with their environment. There is an abundance of video footage available from existing CCTV style cameras, but most do not have the capability to extract valuable analytic data.
New Zealand has uniquely strong anti-surveillance legislation, so the aim of the research is not to identify individuals, but to see patterns of behaviour and extract metadata. The strong privacy culture also drives the need to reduce the need for humans to review video; removing the possibility of bias or abuse entirely.
Intela are experts in machine vision, research was initiated to demonstrate what ‘low hanging fruit’ could be picked off raw video feeds.
We rapidly worked up capability to report:
- People counting
- People direction
- People movement tracking
- Dwell times
- Anonymisation filters
Rich intelligence can be acquired from existing video hardware, without having to compromise privacy.
NOTE: This video is not from New Zealand and can be found on YouTube