Work analyzing system and work analyzing method

1. A work analyzing device in which a processor is caused to perform operations to output analytical information on a work efficiency status of a worker based on second video recordings of working activities of the worker,

wherein the processor is configured to:

generate, based on first video recordings for training and by using machine learning, a posture class estimation model configured to perform a posture match determination;

determine whether the worker is present at a place of working based on the second video recordings;

start measuring a presence time of the worker in response to the worker being at the place of working;

perform a joint position estimation based on the second video recordings to estimate joint positions of the worker;

perform a posture match determination based on results of the joint position estimation and using the posture class estimation model to determine a posture match between a posture of the worker and a predetermined reference body posture;

start measuring a working time of the worker in response to the posture match being determined, the worker being at the place of working, and the worker not working;

stop measuring the working time of the worker in response to the posture match being determined, the worker being at the place of working, and the worker working; and

generate, based on the presence time and the working time, the analytical information on the work efficiency status of the worker.

2. The work analyzing device according to claim 1 , wherein the processor is configured to:

by using the machine learning, create first machine-learned information for the joint position estimation and second machine-learned information for the posture match determination, trained with the first video recordings for training, information records of joint positions of the worker in the first video recordings, and information indicating whether or not to achieve the posture match;

perform the joint position estimation based on the first machine-learned information for the joint position estimation; and

perform the posture match determination based on the results of the joint position estimation and the second machine-learned information for the posture match determination.

3. The work analyzing device according to claim 1 , wherein the processor is configured to:

acquire, based on results of the posture match determination, the working time of the worker during which the worker has been actually working at the place of working, as the analytical information.

4. The work analyzing device according to claim 3 , wherein the processor is configured to:

perform a presence determination to determine whether or not the worker is present at the place of working; and

acquire, based on results of the presence determination, the presence time of the worker during which the worker has been present at the place of working, as the analytical information.

5. The work analyzing device according to claim 3 , wherein the processor is configured to:

acquire a second working time of a skilled worker as a standard time; and

compare the working time of the worker with the standard time, thereby acquiring evaluation information as the analytical information, the evaluation information including an evaluated value of working efficiency of the worker.

6. The work analyzing device according to claim 1 , wherein the processor is configured to:

perform the posture match determination on each of a plurality of workers;

create the analytical information on each worker of the plurality of workers based on results of the posture match determination for each worker; and

perform statistic calculation on analytical information on each worker to generate statistical information on the work efficiency status of the worker.

7. The work analyzing device according to claim 1 , wherein the processor is configured to:

perform the posture match determination to determine the posture match between the posture of the worker and any one of a plurality of predetermined reference body postures; and

generate the analytical information based on the one of the plurality of predetermined reference body postures for which the posture match is determined.

8. The work analyzing device according to claim 1 , wherein the processor is further configured to:

display, on a terminal, an analysis result display screen that includes the analytical information, with the working time and the presence time being superimposed on the analysis result display screen.

9. A work analyzing method for causing a processor to perform operations to output analytical information on a work efficiency status of a worker based on second video recordings of working activities of the worker, the operations comprising:

generating, based on first video recordings for training and by using machine learning, a posture class estimation model configured to perform a posture match determination;

determining whether the worker is present at a place of working based on the second video recordings;

starting measurement of a presence time of the worker in response to the worker being at the place of working;

performing a joint position estimation based on the second video recordings to estimate joint positions of the worker;

performing a posture match determination based on results of the joint position estimation and using the posture class estimation model to determine a posture match between a posture of the worker and a predetermined reference body posture;

starting measurement of a working time of the worker in response to the posture match being determined, the worker being at the place of working, and the worker not working;

stopping measurement of the working time of the worker in response to the posture match being determined, the worker being at the place of working, and the worker working; and

generating, based on the presence time and the working time, the analytical information on the work efficiency status of the worker.