Monitoring the performance of physical exercises

1. A method for monitoring a person performing a physical exercise based on a sequence of image frames showing an exercise activity of the person, the method comprising:

extracting, via a smartphone or tablet, based on the sequence of image frames, for each image frame a set of body key points using a convolutional neural network, the set of body key points being indicative of a posture of the person in the image frame,

wherein the smartphone or tablet is positioned a distance from the person,

wherein the convolutional neural network is trained to recognize the body key points in the image frames;

deriving, based on a subset of the body key points in each image frame, at least one characteristic parameter indicating a progression of a movement of the person, wherein the at least one characteristic parameter is not equal to any coordinate value of the body key points;

automatically detecting a start loop condition by evaluating only a time progression of the at least one characteristic parameter, said start loop condition indicating a transition from a start posture of the person to the movement of the person when performing the physical exercise;

after detecting the start loop condition, analysing the time progression of at least one of the characteristic parameters to evaluate the person's movement and provide feedback to the person; and

automatically detecting an end loop condition by evaluating only the time progression of the at least one of the characteristic parameters, said end loop condition indicating a transition from the person's movement when performing the physical exercise to an intermediate posture.

2. The method according to claim 1 , wherein at least one of the characteristic parameters for a respective image frame is derived from coordinate values of the body key points of the respective image frame.

3. The method according to claim 1 , wherein for each of the image frames, at least one of the characteristic parameters is the Procrustes distance between the subset of body key points in a respective frame and the same subset of body key points in a reference frame.

4. The method according to claim 1 , further comprising detecting the start posture of the person by comparing the person's posture in at least one image frame of the sequence of image frames with at least one predefined criterion.

5. The method according to claim 1 , wherein an image frame in the start loop condition is detected defines the start of the person's exercising activity.

6. The method according to claim 1 , wherein the start loop condition is detected at an image frame in which at least one of the characteristic parameters leaves a predetermined value range and changes with at least a minimum rate of change.

7. The method according to claim 1 , wherein detecting the start loop condition comprises detecting when at least one of the characteristic parameters leaves a predetermined value range corresponding to the person's start posture.

8. The method according to claim 1 , further comprising detecting at least one evaluation point in the person's movement by evaluating the time progression of at least one characteristic parameter indicating the person's movement.

9. The method according to claim 8 , further comprising evaluating the person's posture at the at least one evaluation point.

10. A method for monitoring a person performing a physical exercise based on a sequence of image frames showing an exercise activity of the person, the method comprising:

extracting, via a smartphone or tablet,

based on the sequence of image frames, for each image frame a set of body key points using a convolutional neural network, the set of body key points being indicative of a posture of the person in the image frame,

wherein the smartphone or tablet is positioned a distance from the person,

wherein the convolutional neural network is trained to recognize body key points in the image frames;

deriving, based on a subset of the body key points in each image frame, at least one characteristic parameter indicating a progression of a movement of the person, wherein the at least one characteristic parameter is not equal to any coordinate value of the body key points;

automatically detecting a start loop condition by evaluating only a time progression of at least one of the characteristic parameters, said start loop condition indicating a transition from a start posture of the person to the movement of the person when performing the physical exercise;

detecting at least one evaluation point in the movement of the person by evaluating a time progression of the at least one characteristic parameter;

evaluating a posture of the person in a respective image frame at each evaluation point or in at least one image frame in a respective predefined time interval around each evaluation point; and

automatically detecting an end loop condition by evaluating only the time progression of the at least one of the characteristic parameters, said end loop condition indicating a transition from the person's movement when performing the physical exercise to an intermediate posture.

11. The method according to claim 10 , wherein evaluating the posture of the person comprises comparing the person's posture with a set of predefined conditions.

12. The method according to claim 10 , wherein, based on the result of comparison between the posture of the person and a set of predetermined feedback trigger conditions, feedback is provided to the person.

13. A smartphone or table comprising:

a camera configured to capture a sequence of image frames showing an exercise activity of a person using the smartphone or tablet; and

a controller configured to:

extract a set of body key points using a convolutional neural network for each image frame among the sequence of image frames, the set of body key points being indicative of a posture of the person in each image frame,

wherein the convolutional neural network is trained to recognize body key points in the image frames,

derive, based on a subset of the body key points in each image frame, at least one characteristic parameter indicating a progression of a movement of the person,

automatically detect a start loop condition by only evaluating a time progression of the at least one characteristic parameter, said start loop condition indicating a transition from a start posture of the person to the movement of the person when performing the physical exercise,

analyse the time progression of at least one of the characteristic parameters to evaluate the person's movement and provide feedback to the person, and

automatically detect an end loop condition by evaluating only the time progression of the at least one of the characteristic parameters, said end loop condition indicating a transition from the person's movement when performing the physical exercise to an intermediate posture.

14. A non-transitory computer storage readable medium comprising computer executable program code configured to perform the method according to claim 1 .

15. The method of claim 1 , wherein the smartphone or tablet is a smartphone.

16. The method of claim 10 , wherein the smartphone or tablet is a smartphone.

17. The smartphone or tablet of claim 13 , wherein the smartphone or tablet is a smartphone.

18. A smartphone or tablet comprising:

a camera configured to capture a sequence of image frames showing an exercise activity of a person using the smartphone or tablet; and

a controller configured to:

extract a set of body key points using a convolutional neural network for each image frame among the sequence of image frames, the set of body key points being indicative of a posture of the person in each image frame, wherein the convolutional neural network is trained to recognize body key points in an image frame,

derive, based on a subset of the body key points in each image frame, at least one characteristic parameter indicating a progression of a movement of the person, wherein the at least one characteristic parameter is not equal to any coordinate value of the body key points,

automatically detect a start loop condition by evaluating only a time progression of the at least one the characteristic parameter, said start loop condition indicating a transition from a start posture of the person to the movement of the person when performing the physical exercise,

detect at least one evaluation point in the movement of the person by evaluating a time progression of the at least one characteristic parameter,

evaluate a posture of the person in a respective image frame at each evaluation point or in at least one image frame in a respective predefined time interval around each evaluation point, and

automatically detect an end loop condition by evaluating only the time progression of the at least one the characteristic parameters, said end loop condition indicating a transition from the person's movement when performing the physical exercise to an intermediate posture.