Swim lap counting and timing system and methods for event detection from noisy source data

• US 9,217,634 B1
• Filed: 05/06/2015
• Issued: 12/22/2015
• Est. Priority Date: 05/06/2015
• Status: Active Grant

• Current Assignee: Ocula Corporation
• Original Assignee: Swimpad Corporation
• Inventors: Firmin, Robert L
• Primary Examiner(s): Henson, Mischita
• Assistant Examiner(s): BLOSS, STEPHANIE E
• Application Number:
  US14/705,542
• Time in Patent Office: 230 Days
• Field of Search: 702/150
• US Class Current: 1/1
• CPC Class Codes: G01B 11/14 for measuring distance or clearance between spaced objects or spaced apertures G01B11/26 takes precedence; rangefinders G01C3/00
• CPC Class Codes: G01V 8/10 Detecting, e.g. by using light barriers by reflection from the object G01S17/00
• CPC Class Codes: G04F 10/00 Apparatus for measuring unknown time intervals by electric means
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 18/2193 based on specific statistical tests
• CPC Class Codes: G06F 18/24155 Bayesian classification
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06V 10/764 using classification, e.g. of video objects
• CPC Class Codes: G06V 10/7796 based on specific statistical tests
• CPC Class Codes: G06V 20/42 of sport video content

Abstract

Systems and methods for lap timing and counting in athletic events are disclosed. The systems and methods do not require the athlete to wear a counter/timer, a transmitter, a reflector or another kind of marker. A portable computing device with a sensor, such as a tablet computer with a camera, is positioned in an appropriate location. Data from the sensor is transformed into a time series of data, and one or more learned statistics are calculated in real time as benchmark ambient conditions. The learned statistics are essentially continuously updated and data that indicates irrelevant volatility is excluded. A detection threshold is determined and essentially continuously updated based on the learned statistics, and lap completion is determined based on the threshold. Times, lap counts, and other data are displayed on the portable device in real time.

Claims

1. A detection method, comprising:

gathering data indicating whether an unmodified moving object has passed a defined point using a sensor coupled to a computing device placed in a position to observe the defined point and to observe an event indicating that the moving object has passed the defined point, the computing device being modified or adapted for placement in the position to observe the event;

transforming the data using the computing device from a first form into a second form in which the data can be evaluated as a time series;

evaluating selected data points of the time series of data in real time using the computing device to establish at least one learned statistic that is updated continuously at a defined rate and that represents a set of ambient conditions around the defined point over one or more time spans of undisturbed data, the time spans being selected in real time during said gathering data and being within a time period of said gathering data;

detecting that the event has occurred in real time using the computing device by determining that the time series of data has deviated from the learned statistic in a specified way by more than a defined threshold, the defined threshold determined, at least in part, using the learned statistic, the deviation of the learned statistic in the specified way indicating that the moving object has moved through a detection field of the sensor and has thus passed the defined point; and

calculating and recording or displaying one or both of event count or event timing information based on said detecting.

2. The detection method of claim 1, wherein the event timing information comprises a time duration between either a beginning of the method and the event or a previous event and the event.

3. The detection method of claim 2, further comprising repeating the method to detect multiple events and incrementing the event count with each of the events.

4. The detection method of claim 1, wherein said evaluating further comprises updating the learned statistic with new data values from the time series of data only when the new data values satisfy a maximum acceptable proportional change criterion indicating that a period of non-ambient volatility has passed.

5. The detection method of claim 1, wherein said evaluating further comprises updating the learned statistic with new data values only if the new data values evince a change in the specified way.

6. The detection method of claim 1, wherein said evaluating further comprises evaluating and correcting the level of the learned statistic over time.

7. The detection method of claim 1, further comprising delaying said detecting for a non-zero interval of time such that said detecting occurs only when it is physically possible for the event to occur.

8. The detection method of claim 1, wherein:

the sensor comprises a camera and the computing device comprises a smart phone or tablet computer;

said transforming comprises transforming sensor data into data values derived from light intensity;

the learned statistic comprises a learned ambient mean light intensity value;

the specified way comprises either a positive or a negative change in the data values; and

the defined threshold comprises a multiple of a standard deviation from the learned ambient mean light intensity value that, to within a defined level of reliability or probability, distinguishes between events and non-events.

9. The detection method of claim 1, wherein the unmodified moving object comprises a swimmer in water and the defined point comprises a lap end point.

10. A method of detecting a lap in an athletic event, comprising:

taking data from a camera that is connected to a computing device in a position to observe a defined lap end point, the computing device being modified or adapted for use at the position;

using the computing device, transforming the data from the camera into a time series indicative of light intensity;

with the computing device, calculating and, at a defined rate, continuously updating, a learned ambient mean and a standard deviation in real time using data from the time series taken from undisturbed time spans to exclude from the learned ambient mean and standard deviation periods of volatility that bias detection probabilities;

defining a threshold based on the learned ambient mean and the standard deviation, the threshold indicating that the time series of data has deviated from the learned ambient mean in a specified way to an extent that indicates that an athlete has passed the defined lap end point as observed by the camera, the threshold being continuously updated at a second defined rate;

detecting that an athlete has passed the defined lap end point in real time using the computing device by determining that the threshold has been exceeded; and

displaying one or more of a lap time, a total elapsed time, pacing information or a lap count to the athlete using a screen of the computing device based on said detecting.

11. The method of claim 10, wherein the periods in which ambient conditions are unsettled include one or more of pre-start periods or detected turn periods.

12. The method of claim 10, wherein said calculating further comprises updating the learned ambient mean and the standard deviation with new data values only if the new data values evince a change in the specified way.

13. The detection method of claim 12, wherein said calculating further comprises evaluating and correcting for secular shifts in the levels of the learned ambient mean and the standard deviation.

14. The detection method of claim 13, further comprising delaying said detecting for a non-zero interval of time such that said detecting occurs only when it is physically possible for the athlete to pass the defined lap end point.

15. The method of claim 12, wherein the specified way comprises either a positive change or a negative change in the incident light intensity.

16. The method of claim 10, wherein said transforming comprises transforming the data into units of luma.

17. The method of claim 10, further comprising storing one or both of images or video from the camera.

18. The method of claim 10, wherein the athlete is not required to wear a marker or reflector.

19. A method of detecting laps in swimming, comprising:

taking data from a portable computer placed on the side or bottom of a swimming pool, the portable computer including at least one sensor to generate the data based on ambient conditions in the swimming pool, the portable computer being waterproofed or encased in a waterproof enclosure;

transforming the data into a time series of data points indicative of the ambient conditions;

calculating and, continuously at a defined rate in real time, updating at least one learned statistic by (1) selectively excluding the data points from the learned statistic when those data points indicate a period of non-ambient volatility, (2) including only those of the time series of data points that indicate change in a defined direction or way relevant to the method of detecting laps, (3) periodically checking and adjusting the level of the learned statistic to accommodate longer-term shifts in the ambient conditions;

defining a threshold based on the learned statistic, the threshold indicating that the time series of data has deviated from the learned statistic in a specified way to an extent that indicates that a swimmer has completed a lap, the threshold being continuously updated at a defined rate;

detecting that a swimmer has completed a lap in real time using the portable computer by determining that the threshold has been exceeded, indicating that the swimmer has passed the detection area of the at least one sensor, said detecting being delayed for a non-zero interval of time such that said detecting occurs only when it is physically possible for the swimmer to have completed a lap; and

displaying one or more of a lap time, a total elapsed time, pacing information, or a lap count to the athlete using a screen of the computing device based on said detecting.

20. The method of claim 19, wherein the sensor comprises a camera and said transforming comprises transforming the data into a time series of light intensity values.