Work suitability assessment device and method

Abstract

To provide a work suitability determination apparatus and a method thereof capable of reducing a burden of learning when performing work suitability determination using image recognition.SOLUTION: A method by using a work suitability determination apparatus in a rigging line of a vehicle such as a station wagon uses a wearable camera 6 mounted on a worker to photograph a worker's hand and an object (connector) which is located slightly below in a vehicle vertical direction of a rear glass 2 on a passenger compartment side of a rear gate 1, detects the object in the photographed image with object detection and detects the skeleton of the worker's hand with image recognition, detects the behavior of the worker's hand from time-series changes in the skeleton of the worker's hand, and determines that a predetermined work for the object has been properly performed if the detected object and the detected behavior of the worker's hand substantially match the learned behavior of the worker's hand with respect to the object. When detecting the object, the object is detected based on a location (rear glass).SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a task suitability determination device and method, and more particularly to a task suitability determination device and method for determining whether a task has been performed appropriately when a worker performs a predetermined task with his or her hands on a predetermined object. [Background technology]

[0002] For example, in the manufacturing industry, when a certain task is performed on a certain object, image analysis (image recognition) technology is being used in devices and methods that judge whether the task has been performed properly. For example, the invention described in Patent Document 1 below is a device that captures the content of the task with a wearable camera attached to the worker when it is determined that the worker is performing irregular work that differs from standard work (inspection), and the document describes that whether the workpiece being inspected, imaged by the wearable camera, is a non-defective product is judged by image judgment. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. WO2018 / 062238A1 Summary of the Invention [Problem to be solved by the invention]

[0004] The pass / fail judgment described in Patent Document 1 is performed by comparing a standard image of a pass / fail product with an image of the work captured by a wearable camera, but when the defective conditions and types of work become complex, it becomes necessary to learn about them extensively. Artificial intelligence (hereinafter referred to as AI) with learning capabilities can detect objects from images (videos), detect their positions, and label them for pass / fail judgment. Attempts are being made to use this AI function to judge the suitability of work on the target object.

[0005] In the manufacturing industry, predefined tasks (prescribed tasks) are generally performed by the hands of workers. Therefore, the appropriateness of a prescribed task for an object can be judged, for example, by capturing an image of the worker's hands and the object when the prescribed task is being performed properly with a wearable camera (or a fixed camera), having AI learn the worker's hands and the object in the captured image, and judging whether the prescribed task is appropriate or inappropriate based on whether the worker's hands and the object in the captured image captured during actual work are in the appropriate position and in the appropriate form. There are also cases where the hand and object of a worker when the prescribed task is being performed improperly are learned. The position and shape (form) of the worker's hands themselves in the image are learned.

[0006] However, this task suitability assessment requires, firstly, that the system extracts and learns the worker's hands and the target object from each of a large number of images (still images) cut out from a video. Secondly, when learning and judging the shape of the worker's hand, for example, if the worker changes and the size of the hand changes accordingly, the system must learn about the hand of that size separately. Similarly, if the target object changes and, as a result, the position or shape of the worker's hand changes, re-learning is also required. In other words, when assessing task suitability using image recognition, it is necessary to reduce the burden of learning.

[0007] The present invention has been made in consideration of the above-mentioned problems, and its purpose is to provide a work suitability assessment device and method that can reduce the learning burden when assessing work suitability using image recognition. [Means for solving the problem]

[0008] In order to achieve the above object, a task suitability determination device of the present invention is a task suitability determination device that determines whether a predetermined task has been performed properly when a worker performs a predetermined task with his or her hand on a predetermined object, and includes: a wearable camera capable of capturing an image of an area approximately equal to or larger than the field of vision of the worker; object detection means that detects the object within the image capture area captured by the wearable camera; hand skeleton detection means that detects the skeleton of the worker's hand within the image capture area; motion detection means that detects the behavior of the worker's hand from time-series changes in the object detected by the object detection means and the skeleton of the worker's hand detected by the hand skeleton detection means; and task suitability determination means that determines whether at least the object detected by the object detection means and the behavior of the worker's hand detected by the motion detection means approximately match pre-learned behaviors of the object and the worker's hand, and determines that the predetermined task has been performed properly if they approximately match. The predetermined work is performed at a predetermined timing on a vehicle outfitting line, the motion detection means detects the timing of the worker's hand motion, and the work suitability determination means determines that the predetermined work has been performed properly when the timing of the worker's hand motion detected by the motion detection means substantially coincides with the learned predetermined timing. It is characterized by:

[0009] According to this configuration, the target object and the skeleton of the worker's hand in the imaging area can be detected by image recognition, and the behavior of the hand relative to the target object can be detected by image recognition from the time-series changes of the detected target object and the skeleton of the hand. At this time, the skeleton of the worker's hand during a predetermined task detected by image recognition is approximately the same even if the shape of the worker's hand, such as the size, changes to a certain extent, so that the behavior of the hand detected from the time-series changes can be detected in the same way by learning even if the worker changes. Similarly, even if the target object changes to a certain extent, if the content of the predetermined task is the same, the behavior of the hand relative to the target object can be detected in the same way by learning. Since the target object itself detected by image recognition (object detection) can be reliably detected, including its position information, if the behavior of the target object and the hand of the worker approximately matches the content learned in advance, it can be determined that the predetermined task on the target object has been performed properly. Therefore, re-learning in the task suitability determination when the shape of the worker's hand or the target object changes to a certain extent is reduced, and the learning burden is reduced accordingly. Furthermore, with this configuration, the timing of the worker's hand movements can be taken into account when judging the suitability of the work, so that even in cases where the timing of a specific task is regulated, it can be reliably determined whether the work has been performed properly.

[0010] Another configuration of the present invention is characterized in that, when the specified task is performed at a specified location, the object detection means detects the specified location within an imaging area captured by the wearable camera, and the work suitability determination means determines that the specified task has been performed properly when the learned hand behavior of the worker is detected at the specified location detected by the object detection means.

[0011] According to this configuration, since the predetermined place in the imaging area where the predetermined work is to be performed can be detected by image recognition (object detection, area estimation), it can be determined that the predetermined work has been performed properly when the learned behavior of the worker's hand is detected at the predetermined place. In this case, if the predetermined place where the predetermined work is performed changes, it is only necessary to learn the image recognition content of the changed place, so that the burden of learning, for example, relearning the whole of the worker's hand (position and shape) at the changed place is reduced.

[0014] In a further configuration of the present invention, the work suitability determining means is Regular work The device is characterized by being equipped with a notification means for notifying the user if the transaction is carried out improperly.

[0015] According to this configuration, since a notification is given when a specified operation has been performed improperly, it is possible to prevent the occurrence of defects in mass-produced automobile manufacturing industries, for example.

[0016] Further, a work suitability determination method of the present invention is a work suitability determination method for determining whether a predetermined task has been performed properly when a worker performs a predetermined task with his / her hand on a predetermined object, the method comprising: an object detection step of detecting the object within an imaging area captured by a wearable camera capable of capturing an image of an area substantially equal to or larger than the field of vision of the worker; a hand skeleton detection step of detecting a skeleton of the hand of the worker within the imaging area; a motion detection step of detecting a behavior of the worker's hand relative to the object from a time series change in the object detected in the object detection step and the skeleton of the worker's hand detected in the hand skeleton detection step; and a work suitability determination step of determining whether or not at least the object detected in the object detection step and the behavior of the worker's hand relative to the object detected in the motion detection step substantially match the object and the behavior of the worker's hand relative to the object that have been learned in advance, and determining that the predetermined task has been performed properly if they substantially match. The predetermined work is performed at a predetermined timing on a vehicle outfitting line, the motion detection step detects the timing of the worker's hand motion, and the work suitability determination step determines that the predetermined work has been performed properly if the timing of the worker's hand motion detected by the motion detection means substantially matches the learned predetermined timing. It is characterized by:

[0017] According to this configuration, the target object and the skeleton of the worker's hand in the imaging area can be detected by image recognition, and the behavior of the hand relative to the target object can be detected by image recognition from the time-series changes of the detected target object and the skeleton of the hand. At this time, the skeleton of the worker's hand during a predetermined task detected by image recognition is approximately the same even if the shape of the worker's hand, such as the size, changes to a certain extent, so that the behavior of the hand detected from the time-series changes can be detected in the same way by learning even if the worker changes. Similarly, even if the target object changes to a certain extent, if the content of the predetermined task is the same, the behavior of the hand relative to the target object can be detected in the same way by learning. Since the target object itself detected by image recognition (object detection) can be reliably detected, including its position information, if the behavior of the target object and the worker's hand approximately matches the content learned in advance, it can be determined that the predetermined task on the target object has been performed properly. Therefore, re-learning in the task suitability determination when the shape of the worker's hand or the target object changes to a certain extent is reduced, and the learning burden is reduced accordingly. Effect of the Invention

[0018] As described above, according to the present invention, it is possible to reliably determine whether a specified task on an object has been performed properly using image recognition, and it is also possible to reduce the learning burden, such as the need to re-learn when the object or the worker's hands have changed to some extent. [Brief description of the drawings]

[0019] [Figure 1] 1 is a schematic perspective view showing an embodiment of a mass-produced automobile manufacturing process to which the work suitability determination device and method of the present invention are applied; [Diagram 2] 2 is an explanatory diagram showing an example of an image captured in the manufacturing process of FIG. 1. [Diagram 3] FIG. 11 is an explanatory diagram of detection of the skeleton of a worker's hand. [Figure 4] 2 is a flowchart showing a learning flow in the task suitability determination device and method of FIG. 1. [Diagram 5] 2 is a flowchart showing the procedure for judgment in the task suitability judgment device and method of FIG. 1. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] An embodiment of the work suitability determination device and method of the present invention will be described in detail below with reference to the drawings. FIG. 1 is a schematic perspective view of a mass-produced automobile manufacturing process to which the work suitability determination device and method of this embodiment are applied. This process includes the operation of fitting a connector 3 on the passenger compartment side of a rear gate 1 in an outfitting line for vehicles such as station wagons. The connector 3 is located slightly below the rear window 2 in the vertical direction of the vehicle. The operation of fitting the connector 3 is performed from below the rear gate 1 with the rear gate 1 open, as shown in the figure.

[0021] The work aptitude assessment device of this embodiment is constructed by installing the necessary application software (hereinafter, application) on a personal computer (hereinafter, PC) 5. The main application is an AI-equipped application capable of image recognition. Needless to say, the PC 5 is a computer system with advanced arithmetic processing capabilities, and is configured with a storage device for storing programs and data, and an input / output device for inputting signals from external devices such as cameras and sensors, and outputting signals to external devices such as monitors and speakers. For example, the application for skeleton detection described later may be "unsupervised learning" that detects skeletons simply by reading a video, or may be "supervised learning" that detects skeletons by teaching the positions of the joints of the hand and connecting the joints with lines. Note that "detection" in image recognition is also called "estimation (inference)".

[0022] The work suitability determination device of this embodiment is configured with a wearable camera 6 that can be worn by a worker to obtain images (video) of the connector 3 and the worker's hands, which are the target objects. As a condition for this wearable camera 6, it is desirable that it can capture an area approximately equal to or larger than the field of vision of the worker. In order to properly "fit the connector," it is considered essential that the worker can see at least the "connector" and "the worker's own hands," and it is considered essential that they are captured in order to determine the suitability of the predetermined work. In this embodiment, in order to capture an area equal to or larger than the field of vision of the worker, the wearable camera 6 is attached to the upper surface of the visor provided in front of the worker's hat. Note that the image (video) signal of this wearable camera 6 is transmitted to the PC 5 by a well-known wireless transmission / reception means (not shown). Also, the mounting position of the wearable camera 6 is not limited to the above. Also, the image signal may be transmitted to the PC 5 by a wire.

[0023] FIG. 2 shows an example of an image captured by the wearable camera 6 and read by the PC 5. This image is a schematic diagram of the opened rear gate 1 captured from below, in which a slightly large square in the center of the figure indicates the rear window 2, two small elongated squares slightly above the image indicate the connector 3, and the lines connected to the left and right of the image of the connector 3 indicate the harness 4 (wiring material). When the rear gate 1 is closed, the connector 3 is located on the lower side of the rear window 2, but when the opened rear gate 1 is viewed from below at the rear of the vehicle, it is located above the rear window 2 in the image. Therefore, in the image recognition, for example, it can be detected that the slightly large square part in the center of the image is the rear window 2, and the two small elongated square parts slightly above the image are the connector 3. That is, the rear window 2 is a location (area) necessary for detecting the connector 3 (object), and the connector 3 (object) that is the target can be detected based on the location of the rear window 2. In addition to this information, for example, image recognition can be taught and determined that the connector 3 is "white" with "some parts blue." Such detection can be performed using image recognition AI such as object detection and area estimation, and convolutional networks can be applied, for example. Objects and areas detected in this way are captured, for example, by a rectangular frame that contains them. In other words, the detected object or area exists within the rectangular frame recognized by the image.

[0024] FIG. 3 shows an example of an image of a worker's hand captured by the wearable camera 6 and read by the PC 5, and the hand skeleton detected from the image. The rectangular box shown on the left side of the figure is a schematic representation of the connector 3. In the figure, the outline of the hand in the image is shown by a two-dot chain line, the joint positions are shown by white circles, and the skeleton is shown by solid straight line segments. In recent years, as seen in television broadcasts and telecommunication lines, skeleton detection using image recognition is configured by detecting the joints of the human body and connecting the joints with line segments (also called a skeleton line model). Such skeleton detection using image recognition can detect the skeletons of multiple human bodies that appear overlapping on an image individually, and can also detect the skeletons of human body parts that cannot be detected by conventional object detection, that is, the skeletons of parts that are not clearly shown on an image.

[0025] For example, the skeleton of the hand that pinches the connector 3 with the fingertips shows approximately the same shape for everyone. This skeleton shape is approximately the same even if the size of the hand changes between different workers. In addition, the posture of the worker who fits the connector 3 toward the opened tailgate 1 does not change significantly, so the direction of the hand reflected in the image does not change significantly either. Therefore, if the time series change in the skeleton of the hand is tracked and analyzed during the time when the connector 3 is considered to be fitted, the behavior of the hand when fitting the connector 3 can be recognized and detected. Therefore, the behavior of the hand obtained from the time series change in the skeleton of the hand during the time when the proper connector fitting work is being performed is learned, and when the behavior of the hand that is approximately consistent with the learned behavior of the hand while continuing to detect the connector 3 in the proper position during the actual connector fitting work is detected, it can be determined that the connector fitting work has been performed properly. For example, AI such as a recurrent neural network can be applied to such time series information processing. In addition, since it is time series information, the timing of the behavior of the hand can also be detected and determined. Therefore, when the timing of a predetermined work is specified, it is also possible to determine whether or not the timing is proper.

[0026] FIG. 4 is a flowchart showing the flow of learning performed by the AI-equipped application installed on the PC 5. Note that this flowchart shows supervised learning in which the required number of learnings are taught, but unsupervised learning or deep learning in which the direction of learning is determined by the application itself may also be used. In this learning flow, first, in step S1, learning is performed on the location (area) and target body (object) where a predetermined task is performed from the image (video) captured by the wearable camera 6 and read by the PC 5. In this embodiment, learning is performed with the rear window 2 in the image as the location (reference) and the connector 3 as the target. In addition, the learned location (area) and target body (object) are extracted by the above-mentioned rectangular frame. Note that the image (video) used for learning does not necessarily have to be an image (video) captured by the wearable camera 6, and may be, for example, an image (video) captured of a similar area by another fixed camera.

[0027] Next, the process proceeds to step S2, where the skeleton (joints) is learned from the worker's hand in the image. In this case, as described above, the skeleton lines may be detected by an application and learned, or the joints may be taught and learned by connecting the joints with line segments. Note that the skeleton may be learned from a hand other than that of the actual worker. As described above, considering that workers change hands, it is desirable to learn the skeletons of various hands performing the same task, including artificial images.

[0028] Next, the process proceeds to step S3, where hand behavior is learned from the time series changes in the location and object detected in step S1, and the hand skeleton detected in step S2. The time series changes in the location and object may be the time series changes in the centroid (center of gravity) of the rectangular frame. Also, as described above, hand behavior may be learned by synchronizing the timing of the behavior.

[0029] Next, the process proceeds to step S4, where it is determined whether or not the required number of learning operations have been completed. If the required number of learning operations have been completed, the process returns; if not, the process proceeds to step S1.

[0030] 5 is a flowchart showing the flow of work suitability determination performed by an AI-equipped application installed on the PC 5. In this work suitability determination flow, first, in step S11, a place (area) where a predetermined work is to be performed and a target object (object) are detected from an image (video) captured by the wearable camera 6 and read by the PC 5. In this embodiment, the rear window 2 in the image is detected as the place (reference), and the connector 3 is detected as the target object. In addition, the detected place (area) and target object (object) are extracted using the rectangular frame described above.

[0031] Next, the process proceeds to step S12, where a skeleton is detected from the worker's hand in the image. Instead of directly detecting the skeleton, the joints of the hand may be detected and the skeleton may be detected by connecting the joints with line segments.

[0032] Next, the process proceeds to step S13, where hand behavior is detected from the time series changes in the location and object detected in step S11, and the hand skeleton detected in step S12. The time series changes in the location and object may be the time series changes in the centroid (center of gravity) of the rectangular frame. Also, as described above, hand behavior may be detected by synchronizing the timing of the behavior.

[0033] Next, the process proceeds to step S14, where it is determined whether the location, object, and hand behavior (timing) detected in step S11 and step S13 approximately match the learned content. If they approximately match, the process proceeds to step S15; if not, the process proceeds to step S16.

[0034] In step S15, the process returns after judging that the predetermined work on the target object has been properly performed. In a vehicle outfitting line, for example, the judgment of the proper work may be accompanied by an individual process, such as automatically transporting the vehicle that has been completed to the next process.

[0035] In step S16, the process returns after determining that the predetermined work on the target object was not performed properly. In response to this determination that the work was not performed properly, for example, a notice that the work was not performed properly may be sent via a display, a signal lamp, a buzzer, etc. In addition, for example, in a vehicle outfitting line, individual processing may be performed such that the vehicle is not automatically transported to the next process even when the work is completed.

[0036] According to these calculation processes, a place (rear window 2) and an object (connector 3) are detected from an image of an area approximately equal to or larger than the field of view of the worker, and the behavior of the worker's hand is detected from the time-series changes in the skeleton of the worker's hand shown in the image. If these approximately match the learned content, it is determined that the work of fitting the connector 3 of the rear gate 1 has been performed properly, and if they do not, it is determined that the work has not been performed properly. Among these, the place, i.e., the rear window 2 and the object, i.e., the connector 3, can be reliably detected by object detection, area estimation, etc. As described above, the skeleton of the hand during a predetermined work on the object does not change much even if the worker is different, and the orientation of the hand during a predetermined work performed in a similar posture does not change much either. Therefore, for example, if several skeletons and orientations of the hands and the behavior of the hands due to their time-series changes are learned, the learning burden is much smaller than learning various hand positions and shapes by object detection, etc. Specifically, learning various hand positions and shapes by object detection requires learning for each of a large number of images extracted from a video, specifically, for almost continuous images, but hand skeletons and hand behavior based on time-series changes can be reliably learned and detected even if they are somewhat discrete. In other words, learning hand behavior based on hand skeleton detection and time-series changes has a relatively small learning burden compared to learning hand positions and shapes by object detection. Such learning and detection of hand behavior can be performed in the same way even if the target object changes to a certain extent, for example, if the size of the connector 3 changes slightly.

[0037] In this manner, in the work suitability determination device and method of this embodiment, the target object and the skeleton of the worker's hand in the imaging area captured by the wearable camera 6 can be detected by image recognition, and the behavior of the hand relative to the target object can be detected by image recognition from the time-series changes in the detected target object and skeleton of the hand. At this time, the skeleton of the worker's hand during the specified work detected by image recognition is roughly the same even if the shape of the worker's hand, such as the size, changes to a certain extent, so that the behavior of the hand detected from the time-series changes can be detected in the same way by learning even if the worker changes. Similarly, even if the target object changes to a certain extent, if the content of the specified work is the same, the behavior of the hand relative to the target object can be detected in the same way by learning. Since the target object itself detected by image recognition (object detection) can be reliably detected, including its position information, if the behavior of the target object and the worker's hand approximately matches the content learned in advance, it can be determined that the specified work on the target object has been performed properly. Therefore, the need to re-learn in determining task suitability when the shape of the worker's hand or the target object changes to some extent is reduced, and the learning burden is reduced accordingly.

[0038] Furthermore, when the predetermined work is performed at a predetermined location, the predetermined location in the image capture area where the predetermined work is to be performed can be detected by image recognition (object detection, area estimation), and therefore, when the learned behavior of the worker's hand is detected at the predetermined location, it can be determined that the predetermined work has been performed properly. In this case, when the predetermined location where the predetermined work is performed changes, it is only necessary to learn the image recognition content of the changed location, so that the burden of learning, for example, relearning the entire worker's hand (position and shape) at the changed location is reduced.

[0039] Furthermore, when the predetermined task is performed at a predetermined timing, the timing of the behavior is detected from the time-series changes in the skeleton of the worker's hand, and when the timing of the behavior of the worker's hand substantially matches the learned predetermined timing, it is determined that the predetermined task has been performed properly. This allows the timing of the behavior of the worker's hand to be taken into consideration when determining the appropriateness of the task, so that even when the timing of the predetermined task is stipulated, it is possible to reliably determine whether the task has been performed properly.

[0040] Furthermore, if the above-mentioned predetermined work is performed improperly, a notice to that effect can be given, thereby making it possible to prevent the occurrence of defects in mass-produced automobile manufacturing industries, etc.

[0041] Although the task suitability assessment device and method according to the embodiment have been described above, the present invention is not limited to the configuration described in the above embodiment, and various modifications are possible within the scope of the gist of the present invention. For example, in the above embodiment, a location is learned to assess the suitability of a specific task for an object, and the suitability of the location is also assessed. In addition, the timing of hand behavior can be learned as necessary, and the suitability of the timing of the behavior can also be assessed. On the other hand, the simplest method is to detect an object and assess whether a specific task has been performed properly on the object, so that the location and timing are, so to speak, optional items in assessing task suitability.

[0042] Similarly, if the specified work on the target object has not been performed properly, a notification to that effect is provided; however, even if the improper work is not notified, for example, if the work has been completed on a vehicle body outfitting line but the vehicle is not automatically transported to the next process, it can be recognized that the specified work has not been performed properly.

[0043] In addition, in the above embodiment, only the fitting work of the connector 3 of the vehicle rear gate 1 on the body outfitting line is described in detail, but the work suitability judgment device and method of the present invention are applicable to any work in any industry as long as it is generally in the manufacturing industry. [Explanation of symbols]

[0044] 1 Rear gate 2 Rear window 3 Connector (target body) 5. PC (Personal Computer) 6. Wearable Camera

Claims

1. A task suitability determination device that determines whether a task has been performed appropriately when a worker performs a predetermined task manually on a predetermined object, comprising: A wearable camera capable of capturing an image of an area approximately equal to or larger than the field of view of the worker; An object detection means for detecting the object within an imaging area captured by the wearable camera; a hand skeleton detection means for detecting the skeleton of the hand of the worker within the imaging area; a motion detection means for detecting a motion of the hand of the worker based on a time series change of the object detected by the object detection means and the skeleton of the hand of the worker detected by the hand skeleton detection means; and a task suitability determination means for determining whether or not at least the object detected by the object detection means and the behavior of the operator's hand detected by the motion detection means substantially coincide with previously learned behaviors of the object and the operator's hand, and determining that the predetermined task has been performed properly when the behaviors substantially coincide with each other. The predetermined work is performed at a predetermined timing on a vehicle outfitting line, The motion detection means detects a timing of a hand movement of the worker, The work suitability determination means determines that the specified work has been performed properly when the timing of the worker's hand behavior detected by the motion detection means approximately matches the learned specified timing.

2. The work suitability judgment device as described in Claim 1, characterized in that the timing of the hand behavior is the timing of the hand behavior during the time when the connector fitting operation is being performed.

3. When the predetermined work is performed at a predetermined location, The object detection means detects the predetermined location within an imaging area captured by the wearable camera, The work suitability determination device according to claim 1 or 2, characterized in that the work suitability determination means determines that the specified work has been performed properly when the learned hand behavior of the worker is detected at a specified location detected by the object detection means.

4. 4. The work suitability determination device according to claim 1, wherein the work suitability determination means includes a notification means for notifying a user that the predetermined work has been performed improperly.

5. A method for determining whether a predetermined task has been performed properly when a worker manually performs a predetermined task on a predetermined object, comprising: A target detection step of detecting the target within an imaging area captured by a wearable camera capable of capturing an image of an area approximately equal to or larger than the field of view of the worker; a hand skeleton detection step of detecting a skeleton of the hand of the worker within the imaging area; a motion detection step of detecting a behavior of the hand of the worker relative to the object based on a time series change in the object detected in the object detection step and the skeleton of the hand of the worker detected in the hand skeleton detection step; a task suitability determination step of determining whether or not at least the object detected in the object detection step and the behavior of the operator's hand with respect to the object detected in the motion detection step substantially match the object and the behavior of the operator's hand with respect to the object that have been learned in advance, and determining that the predetermined task has been performed properly if they substantially match, The predetermined work is performed at a predetermined timing on a vehicle outfitting line, In the motion detection step, a timing of a hand movement of the worker is detected; A work suitability determination method characterized in that, in the work suitability determination step, if the timing of the worker's hand behavior detected by the motion detection means approximately matches the learned specified timing, it is determined that the specified work has been performed properly.

6. The method for determining work suitability as described in Claim 5, characterized in that the timing of the hand behavior is the timing of the hand behavior during the time when the connector fitting operation is being performed.

Images