Back to results
US 10,438,473 B2Granted
Activity monitor
AssigneeBuddi Limited
InventorsEwan Findlay, Sara Murray
Life of this patent
Estimating the term…
Abstract
An activity monitor comprising one or more acoustic transducers (100) and a computation component (104) that is arranged to identify events from acoustic signals received by the acoustic transducers (100).
Claims18
An activity monitor for use in a building, the activity monitor comprising:
- one or more acoustic transducers;
- and
- a computation component that is arranged to identify events from acoustic signals received by the one or more acoustic transducers, wherein the computation component is configured:
- to sample a time slice of an incoming acoustic signal among the acoustic signals received by the one or more acoustic transducers;
- to derive a temporal characteristic from the sampled time slice;
- to generate a spectral characteristic of the time slice;
- to combine the temporal characteristic and the spectral characteristic of the time slice to form an acoustic vector;
- and
- wherein the computation component comprises a statistical model-based pattern recognition component arranged to analyse the acoustic vector to identify an event;
- and an event classifier configured to identify indoor actions formed from a sequence of different identified events, wherein a likelihood of an indoor action being performed is derived from a combination of a linear progression of events in a prescribed order;
- and the presence of supplementary events in any order.
8 more elements
A method of monitoring activity, comprising obtaining acoustic signals and computationally identifying events from the acoustic signals by:
- sampling a time slice of an acoustic signal among the acoustic signals;
- deriving a temporal characteristic from the sampled time slice;
- generating a spectral characteristic of the time slice;
- combining the temporal characteristic and the spectral characteristic of the time slice to form an acoustic vector;
- and
- analysing the acoustic vector for identification of an event using a statistical pattern recognition model
- providing an event classifier configured to identify indoor actions formed from a sequence of different identified events, and
- identifying indoor actions wherein a likelihood of an indoor action being performed is derived from a combination of a linear progression of events in a prescribed order;
- and the presence of supplementary events in any order.
6 more elements
A non-transitory computer readable medium having stored thereon instructions that, when executed by a computer, enable the computer it to:
- sample a time slice of an acoustic signal received by an acoustic transducer;
- derive a temporal characteristic from the sampled time slice;
- generate a spectral characteristic of the time slice;
- combine the temporal characteristic and the spectral characteristic of the time slice to form an acoustic vector;
- and
- analyse the acoustic vector for identification of an event using a statistical pattern recognition model
- identify indoor actions formed from a sequence of different identified events using an event classifier, wherein a likelihood of an indoor action being performed is derived from a combination of a linear progression of events in a prescribed order;
- and the presence of supplementary events in any order.
5 more elements
Dependencies are inferred from the claim text.