Exercise assisting apparatus displaying difference from exemplary motions
An exercise assisting apparatus 1 includes: a division unit 312 configured to divide sample motion information 3231 indicating motions of a sample person doing exercise into a plurality of pieces of motion element information 3235 , according to regularity of the motions of the sample person; and a generation unit 314 configured to generate an inference model by causing a pre-trained model 322 to learn time-series antecedent dependency of the plurality of pieces of motion element information, the inference model inferring motions of a target person, based on target motion information 3211 indicating the motions of the target person doing exercise.
1. An exercise assisting apparatus comprising:
at least one memory configured to store instructions; and
at least one processor configured to execute the instructions to:
divide sample motion information indicating motions of a sample person repeating same exercise into a plurality of pieces of motion element information, according to regularity of the motions of the sample person;
generate an inference model by causing a pre-trained model to learn time-series antecedent dependency of the plurality of pieces of motion element information by machine learning, the inference model inferring motions of a target person, based on target motion information indicating the motions of the target person doing exercise;
acquire one or more images of the target person who is repeating same exercise from an image pickup apparatus;
divide motions of the target person in the acquired one or more images into a plurality of pieces of motion element information;
infer a difference between the motions of the target person and exemplary motions by sing the inference model based on the plurality of pieces of motion element information; and
display the inferred difference in a superimposed manner on an image in which the target person appears.
2. The exercise assisting apparatus according to claim 1 , wherein
the at least one processor is configured to execute the instructions to:
display an object and a text message indicating a body part of the target person that is displaced from the exemplary motions, as the inferred difference.
3. The exercise assisting apparatus according to claim 1 , wherein
the inferred difference is inferred based on a difference between a vector indicating the exemplary motions and a vector indicating the motions of the target person.
4. The exercise assisting apparatus according to claim 1 , wherein
the pre-trained model is a model generated through learning time-series antecedent dependency of a plurality of pieces of first motion element information that is the plurality of pieces of motion element information,
the plurality of pieces of first motion element information is generated by dividing, according to the regularity of the motions of the sample person, first sample motion information that is the sample motion information indicating the motions of the sample person doing a first type of exercise,
the at least one processor is configured to execute the instructions to:
divide second sample motion information into a plurality of pieces of second motion element information, the second sample motion information being the sample motion information indicating the motions of the sample person doing a second type of exercise that is different from the first type of exercise, the plurality of pieces of second motion element information being the plurality of pieces of motion element information, and
generate the inference model by causing the pre-trained model to additionally learn time-series antecedent dependency of the plurality of pieces of second motion element information.
5. The exercise assisting apparatus according to claim 2 , wherein
the at least one processor is configured to execute the instructions to:
output notification information in the superimposing manner on the image of the target person, the notification information notifying a body part of the target person that is displaced from the exemplary motions.
6. The exercise assisting apparatus according to claim 4 , wherein
the at least one processor is configured to execute the instructions to display the notification information to optimize the motions of the target person so that the motions of the target person is closer to the exemplary motions.
7. The exercise assisting apparatus according to claim 1 , wherein
the target person is a person who is exercising with a support by a doctor, a nurse, or a physical therapist.
8. An exercise assisting method comprising:
dividing sample motion information indicating motions of a sample person repeating same exercise into a plurality of pieces of motion element information, according to regularity of the motions of the sample person;
generating an inference model by causing a pre-trained model to learn time-series antecedent dependency of the plurality of pieces of motion element information by machine learning, the inference model inferring motions of a target person, based on target motion information indicating the motions of the target person doing exercise;
acquiring one or more images of the target person who is repeating same exercise from an image pickup apparatus;
dividing motions of the target person in the acquired one or more images into a plurality of pieces of motion element information;
inferring a difference between the motions of the target person and exemplary motions by sing the inference model based on the plurality of pieces of motion element information; and
displaying the inferred difference in a superimposed manner on an image in which the target person appears.
9. A non-transitory recording medium storing a computer program causing a computer to execute an exercise assisting method,
the exercise assisting method including:
dividing sample motion information indicating motions of a sample person repeating same exercise into a plurality of pieces of motion element information, according to regularity of the motions of the sample person;
generating an inference model by causing a pre-trained model to learn time-series antecedent dependency of the plurality of pieces of motion element information by machine learning, the inference model inferring motions of a target person, based on target motion information indicating the motions of the target person doing exercise;
acquiring one or more images of the target person who is repeating same exercise from an image pickup apparatus;
dividing motions of the target person in the acquired one or more images into a plurality of pieces of motion element information;
inferring a difference between the motions of the target person and exemplary motions by sing the inference model based on the plurality of pieces of motion element information; and
displaying the inferred difference in a superimposed manner on an image in which the target person appears.