Operation management device

Abstract

To provide a work management device capable of improving work efficiency.SOLUTION: A work management device includes: a detection unit that detects, from an image obtained by capturing a worker working on a workpiece being carried in a predetermined direction, coordinates of a skeleton of the worker in the image; a determination unit that determines actions and positions in the predetermined direction of the worker from changes over time in the coordinates of the skeleton of the worker; a calculation unit that calculates a delay in the work of the worker based on the actions and positions in the predetermined direction of the worker; and a request unit that requests a helper to help the worker in accordance with the delay.SELECTED DRAWING: Figure 5

Description

Technical Field

[0006] , , Another work management device of the present invention includes: a detection unit that detects the coordinates of the worker's skeleton in an image taken of a worker working on a workpiece being transported in a predetermined direction; a determination unit that determines the worker's movement and position in the predetermined direction from the time change of the worker's skeleton coordinates; a calculation unit that calculates the delay of the worker's work based on the worker's movement and position in the predetermined direction; and a request unit that requests assistance from supporters to assist the worker according to the delay. ,

[0009] , , ,

[0001] The present invention relates to a work management device.

Background Art

[0002] For example, Patent Document 1 describes that in a production line of a factory, when an operator presses a push button switch, a request from a work supporter is displayed on a display such as an andon.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, since there are individual differences in the timing when an operator requests support, it is difficult to call a supporter in a timely manner and improve work efficiency.

[0005] Therefore, the present invention has been made in view of the above problems, and an object thereof is to provide a work management device capable of improving work efficiency.

Means for Solving the Problems

[0006] The work management device of the present invention includes a detection unit that detects the coordinates of the skeleton of an operator in an image obtained by imaging an operator working on a workpiece being conveyed in a predetermined direction, a determination unit that determines the movement of the operator and the position in the predetermined direction from the temporal change of the coordinates of the skeleton of the operator, a calculation unit that calculates a delay in the work of the operator based on the movement of the operator and the position in the predetermined direction, and a request unit that requests a supporter to support the operator according to the delay. A control unit that, if the assistant working on the workpiece is not captured in the image, reduces the transport speed of the workpiece in accordance with the delay in the work, and does not reduce the transport speed of the workpiece if the assistant working on the workpiece is captured in the image. It has.

[0009] Another work management device of the present invention includes: a detection unit that detects the coordinates of the worker's skeleton in an image taken of a worker working on a workpiece being transported in a predetermined direction; a determination unit that determines the worker's movement and position in the predetermined direction from the time change of the worker's skeleton coordinates; a calculation unit that calculates the delay of the worker's work based on the worker's movement and position in the predetermined direction; and a request unit that requests assistance from supporters to assist the worker according to the delay.The detection unit detects the coordinates of the skeletons of the worker and the assistant from images taken of the worker and the assistant working on the workpiece, respectively. The discrimination unit determines the worker's movement and position in the predetermined direction from the time change of the worker's skeletal coordinates, and determines the assistant's movement and position in the predetermined direction from the time change of the assistant's skeletal coordinates. The calculation unit calculates the delay of the work based on the worker's movement and position in the predetermined direction, and the assistant's movement and position in the predetermined direction. do .

[0010] In the above-described work management device, the discrimination unit may use a discrimination model that has been trained through supervised learning, which takes the coordinates of the worker's skeleton in the image as input and outputs the worker's movements, and may discriminate the worker's movements based on the time change of the coordinates of the skeleton. [Effects of the Invention]

[0011] According to the present invention, work efficiency can be improved. [Brief explanation of the drawing]

[0012] [Figure 1] Figure 1 is a diagram illustrating an example of a work management system. [Figure 2] Figure 2 shows an example of an image from a camera device that captures a worker during work. [Figure 3] Figure 3 shows an example of the correlation between worker movements and positions. [Figure 4] Figure 4 shows an example of images from a camera device that captured images of workers and support staff during work. [Figure 5] Figure 5 shows an example of how to calculate delays in work performed by workers and support staff. [Figure 6] Figure 6 shows another example of calculating delays in work performed by workers and support staff. [Figure 7] Figure 7 is a flowchart showing an example of the operation of the work management server. [Modes for carrying out the invention]

[0013] (Example of a work evaluation system configuration) Figure 1 is a configuration diagram showing an example of a work management system 9. The work management system 9 is installed, for example, in an assembly plant for automobiles or other products.

[0014] The work management system 9 includes a work management server 1, multiple camera devices 2, multiple display devices 3, and a transport device 4, all of which can communicate with each other via a LAN (Local Area Network) 90. The work management server 1 is an example of a work management device. The work management server 1 manages the work of workers. Camera device 2 and display unit 3 are installed in each worker's workspace. Camera device 2 is an example of an imaging device and captures images of people such as workers. Display unit 3 displays the work status of a worker, such as whether or not assistance is needed. Conveying device 4 is the drive source for the conveyor that transports the workpieces that each worker is working on.

[0015] The work management server 1 has a CPU (Central Processing Unit) 10, ROM (Read Only Memory) 11, RAM (Random Access Memory) 12, HDD (Hard Disk Drive) 13, and a communication port 14. The CPU 10 is electrically connected to the ROM 11, RAM 12, HDD 13, and communication port 14 via a bus 19 so that they can input and output signals to and from each other.

[0016] ROM11 stores the program that drives the CPU10. RAM12 functions as the working memory for the CPU10. Communication port 14 is, for example, a wireless LAN card and handles communication of the CPU10 via LAN90.

[0017] When the CPU 10 reads a program from the ROM 11, it forms a device control unit 100, a skeleton detection unit 101, a work determination unit 102, a delay calculation unit 103, a conveyance speed control unit 104, and a support request unit 105 as software functions. Further, the HDD 13 stores detected skeleton data 130, learned skeleton data 131, work information 132, and a discrimination model 133.

[0018] The device control unit 100 instructs operations to the skeleton detection unit 101, the work determination unit 102, the delay calculation unit 103, the conveyance speed control unit 104, and the support request unit 105 according to the sequence defined in the program. Note that the device control unit 100, the skeleton detection unit 101, the work determination unit 102, the delay calculation unit 103, the conveyance speed control unit 104, and the support request unit 105 are not limited to software and may be realized by hardware such as an ASIC (Application Specified Integrated Circuit).

[0019] The skeleton detection unit 101 is an example of a detection unit. The skeleton detection unit 101 detects the coordinates of the skeleton of a worker in an image from an image of a camera device 2 that images a worker during work on a workpiece being conveyed in a predetermined conveyance direction. The skeleton detection unit 101 receives image data from the camera device 2 via the LAN 90. The skeleton detection unit 101 detects the coordinates of the skeleton of the worker by, for example, image analysis using general-purpose AI (Artificial Intelligence).

[0020] The skeleton detection unit 101 detects the coordinates of the skeleton of the worker from the image of the evaluation worker and stores them in the HDD 13 as detected skeleton data 130. The detected skeleton data 130 shows, for example, the coordinates of the nose, left shoulder, right shoulder, left elbow, right elbow, left hand, right hand, left waist, and right waist of the evaluation worker in the image for each frame number of the image in time series. The coordinates are represented by, for example, X coordinates and Y coordinates (x, y) when the horizontal and vertical directions of the rectangular frame of the image are defined as the X axis and the Y axis, respectively. Note that there is no limitation on the position of the origin.

[0021] The work discrimination unit 102 is an example of a discrimination unit. The work discrimination unit 102 determines the worker's movement and position in the workpiece transport direction from the time change of the coordinates of the worker's skeleton. Examples of worker movements include assembly operations, welding operations, and painting operations. The worker's position is indicated, for example, by the area in which the worker is working, among multiple areas obtained by dividing the image frame of the camera device 2 along a direction approximately perpendicular to the transport direction.

[0022] The work discrimination unit 102 takes the coordinates of the worker's skeleton in the image as input and outputs the worker's movement. Using a discrimination model 133 that has been trained through supervised learning, it discriminates the worker's movement based on the time change of the skeletal coordinates. The coordinates of the exemplary worker's skeleton learned by the discrimination model 133 are stored in the HDD 13 as the learned skeleton data 131. The learned skeleton data 131, like the detected skeleton data 130, shows the coordinates of the exemplary worker's nose, left shoulder, right shoulder, left elbow, right elbow, left hand, right hand, left hip, and right hip in chronological order for each frame number of the image. The discrimination model 133 discriminates the worker's movement based on the degree of similarity between the detected skeleton data 130 and the learned skeleton data 131.

[0023] The discrimination model 133 is stored in the HDD 13 as a set of various arithmetic modules and parameters. In the learning process of the discrimination model 133, the time changes in the coordinates of the skeleton of an exemplary worker during work are used as training data, and annotation is performed on the type of work action. The discrimination model 133 is a neural network that mathematically models the function of the human brain, and the weight coefficients of the activation function of the parts corresponding to neurons are determined based on machine learning. In this way, the work discrimination unit 102 can discriminate the actions of a worker with high accuracy by using the discrimination model 133 that has been trained through supervised learning.

[0024] The delay calculation unit 103 is an example of a calculation unit. The delay calculation unit 103 calculates the delay of the worker's work based on the worker's movements and position in the conveying direction. The further the worker is delayed in their work, the further downstream they are in the conveying direction. Therefore, the delay calculation unit 103 calculates the delay from the correlation between the type of movement and the position in the conveying direction, for example, as will be described later.

[0025] The transport speed control unit 104 is an example of a control unit. The transport speed control unit 104 reduces the transport speed of the workpiece in accordance with the delay in the work. This makes the work easier for the worker and reduces delays.

[0026] The support request unit 105 is an example of a request unit. The support request unit 105 requests support from workers in response to delays in the work. For example, the support request unit 105 outputs the support request to the display unit 3 via LAN. The support worker responds to the support request displayed on the display unit 3 and goes to the work area corresponding to the display unit 3 to support the worker's work.

[0027] For example, if the delay exceeds a threshold, the support request unit 105 determines whether support is necessary based on the work information 132. The work information 132 includes, for example, the skill level and workload of each worker. Alternatively, the support request unit 105 may request support from the worker with the greatest delay among multiple workers.

[0028] In this way, the delay calculation unit 103 calculates the delay in the work based on the worker's movements and position in the transport direction determined from the coordinates of the worker's skeleton, and the support request unit 105 requests support workers from supporters according to the delay in the work. Therefore, it is possible to request support in a timely manner according to the accurate work situation. Consequently, the work management server 1 can improve work efficiency.

[0029] The transport speed control unit 104 does not reduce the transport speed of the workpiece when an assistant working on the workpiece is captured in the image of the camera device 2. As a result, after the assistant joins the work, delays are reduced due to the assistant's work, and the decrease in work efficiency is suppressed by maintaining the transport speed of the workpiece.

[0030] (Example image of a worker at work) Figure 2 shows an example of image G1 from camera device 2, which captures worker Ha during work. Camera device 2 captures worker Ha from the front. In image G1, the horizontal direction is the X direction and the vertical direction is the Y direction.

[0031] Image G1 shows worker Ha performing an assembly operation, attaching part C to workpiece W. In this example, workpiece W is an automobile, but it is not limited to this. Workpiece W is placed on a conveyor R and transported in a transport direction that coincides with the X direction in Figure 2.

[0032] The skeleton detection unit 101 detects the coordinates of worker Ha's skeleton from image G1 using general-purpose AI image analysis and stores them in chronological order as detected skeleton data 130. Examples of the skeleton coordinates include the position of worker Ha's nose P0, left shoulder P23, right shoulder P13, left elbow P22, right elbow P12, left hand P21, right hand P11, left hip P32, and right hip P31.

[0033] The work discrimination unit 102 determines the worker Ha's actions based on the learned skeleton data 131, using the time changes in the coordinates of each part of the skeleton shown in the detected skeleton data 130. The work discrimination unit 102 compares the coordinates of the skeleton for various exemplary actions shown in the learned skeleton data 131 with the coordinates of each part of the skeleton shown in the detected skeleton data 130 in a time series. The work discrimination unit 102 determines that the time changes in the positions P0, P11-P13, P21-P23, and P31-P33 of each part of worker Ha are similar to the time changes in the positions P0, P11-P13, P21-P23, and P31-P33 of each part during exemplary assembly operations, and determines worker Ha's assembly operation of part C.

[0034] Furthermore, the work determination unit 102 determines the position of worker Ha in the transport direction from the coordinates of the skeleton among areas #1 to #4 in image G1. Areas #1 to #4 are set by substantially dividing the region in image G1 into four equal parts along the Y direction perpendicular to the transport direction. The work determination unit 102 determines that the position of worker Ha is in area #3 because the positions P0, P11 to P13, P21 to P23, and P31 to P33 of each part of worker Ha are within area #3. The work determination unit 102 does not necessarily need to use the coordinates of all skeletons to determine the position; for example, it may determine the position of worker Ha as area #1 to #4 where the coordinates of the skeleton of worker Ha's torso (P13, P23, 31, P32) are located.

[0035] The delay calculation unit 103 calculates the delay of the work based on the movements and position of worker Ha. For example, the delay calculation unit 103 calculates the delay from the correlation between the movements and position of worker Ha.

[0036] Figure 3 shows an example of the correlation between worker Ha's actions and position. As an example, worker Ha performs actions #1 to #4 in this order. Figure 3 shows worker Ha's actions #1 to #4 and areas #1 to #4 in chronological order when there is no delay.

[0037] If worker Ha performs actions #1 to #4 sequentially in areas #1 to #4, the work delay is always 0. However, if worker Ha's actions #1 to #4 are performed in an area #1 to #4 downstream from the corresponding area #1 to #4, a work delay is determined to have occurred. For example, if worker Ha performs action #2 in area #3, the work delay is 1 (=3-2), and if worker Ha performs action #1 in area #3, the work delay is 2 (=3-1). Also, if worker Ha performs action #2 in area #4, the work delay is 2 (=4-2), and if worker Ha performs action #1 in area #4, the work delay is 3 (=4-1).

[0038] Therefore, the delay calculation unit 103 calculates the work delay (ji) from worker Ha's action #i and area #j (i, j: positive integers) if j > i. The delay calculation unit 103 calculates the work delay as 0 if j ≤ i. The support request unit 105 requests support from a support worker if the work delay is greater than the threshold (>0).

[0039] (Example images of workers and support staff during the work) Figure 4 shows an example of image G2 from camera device 2, which captures images of worker Ha and assistant Hb during work. Camera device 2 captures images of worker Ha and assistant Hb from the front. In Figure 4, components common to Figure 2 are denoted by the same reference numerals, and their explanations are omitted. As an example, assistant Hb is performing an assembly operation in area #4 downstream of worker Ha, assembling part D onto workpiece W. It is assumed that the assembly of part D is a subsequent process after the assembly of part C.

[0040] The skeleton detection unit 101 detects the coordinates of the skeletons of worker Ha and assistant Hb from image G2. Although the symbols are omitted, the coordinates of assistant Hb's skeleton are also detected, including the position of the nose P0, the position of the left shoulder P23, the position of the right shoulder P13, the position of the left elbow P22, the position of the right elbow P12, the position of the left hand P21, the position of the right hand P11, the position of the left hip P32, and the position of the right hip P31. The coordinates of assistant Hb's skeleton are stored as detected skeleton data 130.

[0041] The work determination unit 102 determines worker Ha's movement and position in the transport direction from the time change of worker Ha's skeletal coordinates, and determines supporter Hb's movement and position in the transport direction from the time change of supporter Hb's skeletal coordinates. The means for determining supporter Hb's movement and position is the same as in the case of worker Ha described above. The work determination unit 102 determines worker Ha's assembly operation of part C and area #3 from worker Ha's skeletal coordinates, and determines supporter Hb's assembly operation of part D and area #4 from supporter Hb's skeletal coordinates.

[0042] The delay calculation unit 103 calculates the delay based on the movements and position in the transport direction of worker Ha, and the movements and position in the transport direction of support worker Hb. As an example, the delay calculation unit 103 detects the movements of the person working downstream in the transport direction and the position of the person performing the latest process among worker Ha and support worker Hb, and uses them to calculate the delay. In this example, the delay is calculated based on the assembly operation of part D by support worker Hb and area #4.

[0043] (Example of calculating work delays) Figure 5 shows an example of calculating the delay in work performed by worker Ha and assistant Hb. Symbols G1a to G1d schematically show images from camera device 2 in time series (see arrows). In Figure 5, components common to Figure 4 are given the same symbols, and their explanations are omitted.

[0044] As indicated by the symbol G1a, worker Ha is performing operation #1 in area #3. The delay in this operation is 2 (=3-1) as calculated using the method described above. If the delay threshold for requesting assistance is set to 1, then since the delay exceeds the threshold, a request for assistance is displayed on display 3.

[0045] Next, as indicated by the symbol G1b, assistant Hb arrives at the work area as requested and performs the subsequent operation #2 in area #4, which is downstream of worker Ha. Meanwhile, worker Ha continues operation #1 in area #3. The delay in this work is calculated as 2 (=4-2), based on the delay in area #4, which is downstream of areas #3 and #4, and the delay in the subsequent operation #2, which is one of operations #1 and #2.

[0046] Next, as indicated by the symbol G1c, supporter Hb completes action #2 and performs action #3, while worker Ha completes action #1 and performs action #3 together with supporter Hb. The delay in this work is calculated as 1 (=4-3) from the downstream area #4 and action #3 of the working areas #3 and #4.

[0047] Next, as indicated by the symbol G1d, both worker Ha and support worker Hb have completed operation #4 and are now performing operation #4. The delay in this operation is calculated to be 0 (=4-4).

[0048] Figure 6 shows another example of calculating the work delays of worker Ha and assistant Hb. Symbols G2a to G2c schematically show images from camera device 2 in time series (see arrows). In Figure 6, components common to Figure 5 are given the same symbols, and their explanations are omitted.

[0049] As indicated by the symbol G2a, worker Ha is performing operation #2 in area #4. The delay in this operation is 2 (=4-2) as calculated using the method described above. If we set the delay threshold for requesting assistance to 1, then since the delay exceeds the threshold, a request for assistance is displayed on display 3.

[0050] Next, as indicated by the symbol G2b, assistant Hb arrives at the work area as requested and performs the subsequent operation #3 in area #3, which is upstream of worker Ha. Meanwhile, worker Ha continues operation #2 in area #4. The delay in this operation is calculated as 1 (=4-3), based on the delay in area #4, which is downstream of areas #3 and #4, and the delay in the subsequent operation #3, which is one of operations #1 and #2.

[0051] Next, as indicated by the symbol G2c, both worker Ha and support worker Hb are performing action #4 in area #4. The delay in this work is calculated to be 0 (=4-4) from the downstream area #4 and action #4 of the areas #3 and #4 where work is being performed.

[0052] In this way, the delay calculation unit 103 calculates the work delay based on the movements and position in the transport direction of worker Ha, and the movements and position in the transport direction of support worker Hb. Therefore, the work management server 1 can indicate the degree to which the delay has been resolved due to support worker Hb joining the work. At this time, the delay calculation unit 103 may output the delay to the display unit 3. Note that the method for calculating the work delays of worker Ha and support worker Hb is not limited to this example.

[0053] (Operation of the work management server) Figure 7 is a flowchart illustrating an example of the operation of the work management server 1. This operation is performed in parallel for each of the multiple workers. First, the skeleton detection unit 101 acquires an image from the camera device 2 via the LAN 90 (step St1). Next, the skeleton detection unit 101 detects the coordinates of the skeletons of one or more people in the image (step St2).

[0054] The work determination unit 102 determines whether or not there is a helper in the image based on the skeletal detection result (step St3). If there is no helper (No. in step St3), the work determination unit 102 determines the worker's movement and position in the transport direction (step St4). Next, the delay calculation unit 103 calculates the work delay based on the worker's movement and position in the transport direction (step St5).

[0055] Next, the support request unit 105 compares the work delay with the threshold TH_L (step St6). If work delay ≤ TH_L is true (No in step St6), the device control unit 100 determines whether all work has been completed based on the worker's actions (step St13). If work remains (No in step St13), each operation from step St1 onwards is performed again, and if all work is completed (Yes in step St13), the operation of the work management server 1 ends.

[0056] If the condition "Work delay > TH_L" is met (Yes in step St6), the support request unit 105 determines whether support is needed based on the work information 132 (step St7). At this time, the support request unit 105 calculates the degree of need for support using a predetermined calculation formula based on the worker's skill level and workload included in the work information 132. If the degree is above a certain level, it determines that support is needed (Yes in step St7), and if it is below a certain level, it determines that support is not needed (No in step St7).

[0057] If no assistance is needed (No in step St7), the operation in step St13 is performed. If assistance is needed (Yes in step St7), the assistance request unit 105 requests assistance from an assistant by displaying on the display unit 3 that assistance is needed (step St8). The assistant sees the display on the display unit 3 and heads to the work area.

[0058] Next, the transport speed control unit 104 compares the delay in the work with a threshold TH_H (step St9). Threshold TH_H is greater than the threshold TH_L mentioned above. If the delay in the work ≤ TH_H is true (No in step St9), the operation in step St13 is performed. Also, if the delay in the work > TH_H is true (Yes in step St9), the transport speed control unit 104 instructs the transport device 4 to reduce the transport speed of the conveyor (step St10). After that, the operation in step St13 is performed.

[0059] The symbol Gv indicates the relationship between the work delay (horizontal axis) and the conveyor speed (vertical axis). When the work delay is within the range of 0 to the threshold TH_H, the conveyor speed control unit 104 maintains the conveyor speed at a constant value K. When the work delay exceeds the threshold TH_L, the support request unit 105 requests support, and when the work delay exceeds the threshold TH_H, the conveyor speed control unit 104 controls the conveyor speed to a value lower than the constant value K. In this case, for example, the correlation between the work delay and the conveyor speed is linear. Therefore, if the work delay is increasing before support arrives at the workplace, the conveyor speed control unit 104 can suppress the work delay by reducing the conveyor speed.

[0060] Furthermore, the transport speed control unit 104 may change the timing of the start of the decrease in transport speed according to the skill level of the worker. In this case, for example, the transport speed control unit 104 adjusts the threshold TH_H according to the skill level of the work information 132, as indicated by arrow m. The transport speed control unit 104 lowers the threshold TH_H the lower the skill level. The dotted line is an example of the correlation between work delay and transport speed when the skill level is low. As a result, the transport speed decreases more for less experienced workers, thus more effectively suppressing work delays.

[0061] Furthermore, if there is a support worker (Yes in step St3), the work determination unit 102 determines the worker's movements and position in the transport direction, and the support worker's movements and position in the transport direction (step St11). Next, the delay calculation unit 103 calculates the work delay based on the worker's movements and position in the transport direction, and the support worker's movements and position in the transport direction (step St12). After that, the operation in step St13 is performed. In this way, if there is a support worker (Yes in step St3), the transport speed control unit 104 does not reduce the transport speed of the workpiece, thus suppressing the decrease in work efficiency due to the decrease in transport speed. The work management server 1 operates in this manner.

[0062] The embodiments described above are preferred examples of the present invention. However, the invention is not limited thereto, and various modifications are possible without departing from the spirit of the invention. [Explanation of symbols]

[0063] 1 Work management server (work management device), 10 CPU, 101 Skeleton detection unit (detection unit), 102 Work discrimination unit (discrimination unit), 103 Delay calculation unit (calculation unit), 104 Transport speed control unit (control unit), 105 Support request unit (request unit), 130 Detected skeleton data, 131 Learning skeleton data, 133 Discrimination model

Claims

1. A detection unit that detects the coordinates of the worker's skeleton in an image taken of a worker working on a workpiece being transported in a predetermined direction, A discrimination unit that determines the worker's movement and position in the predetermined direction from the time change of the coordinates of the worker's skeleton, A calculation unit that calculates the delay in the worker's work based on the worker's movements and position in the predetermined direction, A request unit that requests support from the support unit to assist the worker in response to the aforementioned delay, The control unit has the following features: if the assistant working on the workpiece is not captured in the image, it reduces the transport speed of the workpiece in accordance with the delay in the work; and if the assistant working on the workpiece is captured in the image, it does not reduce the transport speed of the workpiece. Work management device.

2. A detection unit that detects the coordinates of the worker's skeleton in an image taken of a worker working on a workpiece being transported in a predetermined direction, A discrimination unit that determines the worker's movement and position in the predetermined direction from the time change of the coordinates of the worker's skeleton, A calculation unit that calculates the delay in the worker's work based on the worker's movements and position in the predetermined direction, It has a request unit that requests support from the support unit to assist the worker in response to the aforementioned delay, The detection unit detects the coordinates of the skeletons of the worker and the assistant from images taken of the worker and the assistant working on the workpiece, respectively. The discrimination unit determines the worker's movements and position in the predetermined direction from the time change of the worker's skeletal coordinates, and determines the supporter's movements and position in the predetermined direction from the time change of the supporter's skeletal coordinates. The calculation unit calculates the delay of the work based on the worker's movements and position in the predetermined direction, and the assistant's movements and position in the predetermined direction. Work management device.

3. The discrimination unit uses a supervised learning-based discrimination model that takes the coordinates of the worker's skeleton in the image as input and outputs the worker's movements, and discriminates the worker's movements based on the time change of the coordinates of the skeleton. The work management device according to claim 1 or 2.

Images