Efficient activity classification from motion inputs
First Claim
1. A method of classifying an activity being performed by a moving body, comprising:
- receiving a stream of samples of acceleration data from a three-axis accelerometer worn by the moving body, each sample comprising a value for each of the three axes;
establishing a sequence of sampling periods, each having a plurality of samples from the stream;
maintaining a mean of at least a portion of the values for each axis;
sorting the samples received during each of the sampling periods into an activity histogram for that sampling period, the activity histogram comprising a plurality of bins, the sorting comprisingfor each of the values of that sample, characterizing a respective difference between that value, and the mean for values of that axis,and selecting a bin of the plurality of bins of the activity histogram for each sample according to the characterized differences for the values of that sample; and
outputting an activity classification by comparing the activity histogram with activity histograms associated with classified activities.
1 Assignment
0 Petitions
Accused Products
Abstract
A device worn by a human performing a method using accelerometer data to classify human activities is disclosed. The method uses memory and computation efficiently. A stream of samples is divided into a sequence of sampling periods; each sample has an acceleration value for each axis (e.g., a 3-D accelerometer). Standard deviation of the values for each axis during each sampling period are calculated. A running sum of each axis'"'"' values can be maintained sample-by-sample. Each value is sorted into a bin of a histogram, by quantifying a deviance from a respective mean in standard deviations. A standard deviation produced from samples of a previous period can be used. The histogram is compared with histograms associated with particular activities and a classification output can be produced. Classification outputs from multiple sampling periods can be used for voting. A threshold amount of activity can be required to begin activity classification.
3 Citations
29 Claims
-
1. A method of classifying an activity being performed by a moving body, comprising:
-
receiving a stream of samples of acceleration data from a three-axis accelerometer worn by the moving body, each sample comprising a value for each of the three axes; establishing a sequence of sampling periods, each having a plurality of samples from the stream; maintaining a mean of at least a portion of the values for each axis; sorting the samples received during each of the sampling periods into an activity histogram for that sampling period, the activity histogram comprising a plurality of bins, the sorting comprising for each of the values of that sample, characterizing a respective difference between that value, and the mean for values of that axis, and selecting a bin of the plurality of bins of the activity histogram for each sample according to the characterized differences for the values of that sample; and outputting an activity classification by comparing the activity histogram with activity histograms associated with classified activities. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19)
-
-
20. A device to be worn by a moving body to produce classifications of activities being performed by that creature, comprising:
-
a multi-axis accelerometer; a non-transitory memory; a processor coupled to receive samples, at a sampling rate, of acceleration sensed by the multi-axis accelerometer, each sample comprising a value for each sampled axis, the processor configured to perform a method comprising; maintaining, for each axis of accelerometer data, a respective current sum and a square of the current sum of the values received during a sampling period for that axis, estimating a respective standard deviation for the values of each axis during each sampling period; sorting each sample into one bin of a histogram having n^3 bins, by sorting the per-axis values of that sample into one of n categories, the sorting comprising characterizing an amount of difference between each value of that sample with a mean of previously-received sample values for that axis, the characterizing expressed as a factor of the standard deviation of the values for that axis; incrementing a per-bin count for each sample sorted into that bin; comparing the histogram with a set of histograms that have been determined to be indicative of different activities, and outputting a classification of the activity being performed during the sampling period based on a result of the comparing. - View Dependent Claims (21, 22, 23, 24, 25, 26, 27, 28)
-
-
29. A non-transitory machine readable medium, the medium storing machine executable instructions for configuring a machine to perform a method of classifying an activity being performed by a moving body, comprising:
-
receiving a stream of samples of acceleration data from a three-axis accelerometer worn by the moving body, each sample comprising a value for each of the three axes; establishing a sequence of sampling periods, each having a plurality of samples from the stream; maintaining a mean of at least a portion of the values for each axis; sorting the samples received during each of the sampling periods into an activity histogram for that sampling period, the activity histogram comprising a plurality of bins, the sorting comprising for each of the values of that sample, characterizing a respective difference between that value, and the mean for values of that axis, and selecting a bin of the plurality of bins of the activity histogram for each sample according to the characterized differences for the values of that sample; and outputting an activity classification by comparing the activity histogram with activity histograms.
-
Specification