US 12,453,898 B2
Increasing accuracy in workout autodetection systems and methods
Stephen John Black, Portland, OR (US)
Assigned to adidas AG, Herzogenaurach (DE)
Filed by adidas AG, Herzogenaurach (DE)
Filed on Oct. 9, 2023, as Appl. No. 18/483,357.
Application 18/483,357 is a continuation of application No. 17/352,709, filed on Jun. 21, 2021, granted, now 11,779,810.
Application 17/352,709 is a continuation of application No. 15/889,407, filed on Feb. 6, 2018, granted, now 11,040,246, issued on Jun. 22, 2021.
Prior Publication US 2024/0042280 A1, Feb. 8, 2024
Int. Cl. G06F 3/048 (2013.01); A63B 24/00 (2006.01)
CPC A63B 24/0062 (2013.01) [A63B 24/0006 (2013.01); A63B 24/0021 (2013.01); G06F 3/048 (2013.01); A63B 2024/0056 (2013.01); A63B 2024/0065 (2013.01); A63B 2024/0068 (2013.01); A63B 2024/0071 (2013.01)] 15 Claims
OG exemplary drawing
 
1. A method of identifying an activity of an individual as a workout, the method comprising:
receiving, via a motion sensor module, a stream of timestamped motion data associated with the activity, the timestamped motion data comprising a timestamp and a corresponding motion data value;
converting, at one or more processors, the stream of timestamped motion data into a motion segment data stream, the motion segment data stream comprising a plurality of duration values and corresponding motion data values;
converting, at the one or more processors, the motion segment data stream into a plurality of timestamped data buckets, each defined by a timespan and each comprising a motion classification and a corresponding motion classification density, the motion classification density indicating a percentage of motion data from the timespan that is categorized within the motion classification;
determining, at the one or more processors, whether a first timestamped data bucket of the plurality of timestamped data buckets comprises one or more motion classification densities that together meet a candidate start density threshold;
analyzing, at the one or more processors, a sequence of timestamped data buckets of the plurality of timestamped data brackets to determine whether an average active density for the sequence meets a predetermined threshold; and
categorizing, at the one or more processors, the activity as a workout based on the average active density for the sequence meeting the predetermined threshold;
the method further comprising:
storing, in computer memory, the first timestamped data bucket even though the one or more motion classification densities do not meet the candidate start density threshold, in response to the candidate start density threshold being met by data of a previous timestamped data bucket of the plurality of timestamped data buckets, or
not storing, in the computer memory, the first timestamped data bucket in response to the one or more motion classification densities not meeting the candidate start density threshold,
wherein implementation of the method improves the performance and efficiency of one or more devices implementing the method.