Pose-based worker syncope detection method

Abstract

The invention discloses a worker syncope detection method based on poses. The method comprises the following steps: acquiring image information of a target area; using a preset pose detection model to detect and track the pose of a person in the target area according to the acquired image information; wherein the pose detection model is established on the basis of an improved YOLO-POSE structure, a weight is introduced into a multi-level feature map in the improved YOLO-POSE structure through BiFPN, and feature fusion on different scales is realized through the weight; analyzing the moving distance of the pose key point of the target person in a preset time period, and judging whether the pose key point is in a static state or not according to the moving distance; and judging whether the target person is in a syncope state or not according to the number of the pose key points in the static state. According to the method, the detection precision of the human body key points in the working process of workers can be improved, meanwhile, a strict alarm filtering mechanism is achieved, and good applicability and generalization ability are achieved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of target detection, and in particular, to a method for detecting worker syncope based on pose. Background Art

[0002] The workshop where workers perform painting operations is a closed environment. If the working time inside is too long, there may be situations where personnel fall or faint. For such emergencies, a warning needs to be issued in a timely manner to avoid personnel losses caused by untimely treatment. According to industrial research, personnel fall detection technology in the production environment can help prevent accidents and improve work safety. In the early days, the abnormal state of workers was mainly judged by the personnel inside the workshop or behind the camera, which relied on manual labor. At present, the detection methods for the abnormal state of workshop workers can be divided into two types: one is to judge the acceleration of the human body's center of mass through sensors, and the other is to perform real-time processing on videos through computer vision, and judge whether the posture of the worker in the image is abnormal by judging the posture of the worker.

[0003] Due to problems such as high equipment costs, low recognition accuracy, and limited usage scenarios in sensor technology, the method of computer vision is more widely used. One of the most common fall detection means in computer vision is OpenPose, but this method is limited by complex environments and cannot detect occluded postures well, and there is no detection frame, so it cannot achieve target tracking. In addition, some methods in the prior art also adopt technical means of target detection to directly locate and distinguish standing and falling postures. These methods have relatively single functions and high requirements for data, are difficult to apply to video data of personnel with different perspectives, and have the technical problem of low accuracy. Summary of the Invention

[0004] In view of the above technical problems, at least one embodiment of the present invention provides a method for detecting worker syncope based on pose. By introducing a deep learning algorithm with an improved YOLO-POSE structure, the detection accuracy of human key points during the worker's operation is greatly improved. At the same time, a more scientific and reasonable alarm filtering mechanism is adopted to reduce the false alarm probability, and it has good applicability and generalization ability.

[0005] In some optional embodiments, the method includes the following steps:

[0006] Obtain the image information of the target area;

[0007] Using a preset pose detection model, detect and track the pose of a person in a target area according to the acquired image information; wherein, the pose detection model is established based on an improved YOLO-POSE structure, and in the improved YOLO-POSE structure, multi-level feature maps introduce weights through BiFPN, and feature fusion at different scales is achieved through the weights;

[0008] Analyze the moving distance of the pose key points of the target person within a preset time period, and judge whether the pose key points are in a static state based on the moving distance;

[0009] Judge whether the target person is in a fainting state according to the number of pose key points in a static state.

[0010] In some alternative embodiments, the weights are used to represent the importance of features for affecting the detection result.

[0011] In some alternative embodiments, the analyzing the moving distance of the pose key points of the target person within a preset time period includes: calculating the Euclidean distance of each pose key point on the picture before and after several frames according to the following formula:

[0012]

[0013] wherein, P i,x and P i,y are the horizontal and vertical coordinates of the i-th historical frame pose key point, and C j,x and C j,y are the horizontal and vertical coordinates of the j-th current frame pose key point.

[0014] In some alternative embodiments, the judging whether the pose key points are in a static state based on the moving distance includes:

[0015] Compare the moving distance of each pose key point with its corresponding distance threshold;

[0016] If the moving distance is lower than the corresponding distance threshold, judge that the pose key point is in a static state;

[0017] If the moving distance is higher than or equal to the corresponding distance threshold, judge that the pose key point is in a non-static state.

[0018] In some alternative embodiments, the judging whether the target person is in a fainting state according to the number of pose key points in a static state includes:

[0019] When the number of pose key points in a static state exceeds a preset number threshold, judge that the target person is in a fainting state.

[0020] In some alternative embodiments, the method further includes: when the target person is in a fainting state, determining whether to send an alarm message.

[0021] In some alternative embodiments, the step of determining whether to send an alarm message when the target person is in a fainting state includes:

[0022] Counting the number of image frames in which the target person is in a fainting state;

[0023] When the target person first appears in a fainting state and the number of image frames in which the target person is in a fainting state exceeds a preset frame number threshold, sending an alarm message.

[0024] At least one embodiment of the present invention further provides a pose-based worker fainting detection device, which is characterized by including:

[0025] An image information acquisition module, configured to acquire image information of a target area;

[0026] A pose detection and tracking module, configured to use a preset pose detection model to detect and track the pose of a person in the target area according to the acquired image information; wherein, the pose detection model is established based on an improved YOLO-POSE structure, and in the improved YOLO-POSE structure, multi-level feature maps introduce weights through BiFPN, and feature fusion at different scales is achieved through the weights;

[0027] A static state determination module, configured to analyze the moving distance of the pose key points of the target person within a preset time period, and determine whether the pose key points are in a static state through the moving distance;

[0028] A fainting state determination module, configured to determine whether the target person is in a fainting state according to the number of pose key points in a static state.

[0029] At least one embodiment of the present invention further provides an electronic device, which is characterized by including:

[0030] At least one processor; and,

[0031] A memory communicatively connected to the at least one processor; wherein,

[0032] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the pose-based worker fainting detection method as described above.

[0033] At least one embodiment of the present invention further provides a computer-readable storage medium storing a computer program, which when executed by a processor implements the above-mentioned pose-based worker syncope detection method.

[0034] At least one embodiment of the present invention further provides a computer program product including a computer program, which when executed by a processor implements the steps of the above-mentioned pose-based worker syncope detection method.

[0035] Compared with the prior art, the pose-based worker syncope detection method provided by the embodiment of the present invention has the following

[0036] Advantages:

[0037] 1) By introducing a complex feature fusion mechanism, the method improves the original YOLO-POSE structure, which can not only be used to warn personnel who have not moved for a long time, but also accurately detect and track personnel.

[0038] 2) The method greatly improves the detection accuracy of human key points during the worker operation process, and at the same time designs a relatively strict alarm filtering mechanism, having good applicability and generalization ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] One or more embodiments are exemplarily illustrated by the pictures in the corresponding drawings, and these exemplary illustrations do not limit the embodiments.

[0040] Figure 1 is a flowchart of the steps of the pose-based worker syncope detection method adopted by the embodiment of the present invention;

[0041] Figure 2 is a schematic diagram of the perspective of the security camera in the workshop provided by the embodiment of the present invention;

[0042] Figure 3 is a flowchart of the processing steps of the pose detection and tracking algorithm adopted by the embodiment of the present invention;

[0043] Figure 4 is a schematic diagram of the improved pose detection model structure adopted by the embodiment of the present invention;

[0044] Figure 5 is a BiFPN feature fusion structure diagram of the three-layer feature map adopted by the embodiment of the present invention;

[0045] Figure 6 is a comparison chart of the indexes of the original algorithm and the improved algorithm provided by the embodiment of the present invention;

[0046] Figure 7It is a schematic diagram of the detection result of the original YOLOv8-POSE algorithm provided by the embodiment of the present invention;

[0047] Figure 8 It is a schematic diagram of the detection result of the improved YOLOv8-POSE algorithm provided by the embodiment of the present invention;

[0048] Figure 9 It is a schematic diagram of the detection result of the video stream in the production environment provided by the embodiment of the present invention. Detailed implementation manners

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, those of ordinary skill in the art can understand that in the embodiments of the present invention, many technical details are provided to help readers better understand the present invention. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed by the present invention can still be implemented. The following division of each embodiment is for convenience of description and should not constitute any limitation to the specific implementation manner of the present invention. Each embodiment can be combined and cross-referenced with each other on the premise of not conflicting.

[0050] As described above, to solve the technical problems existing in the prior art, the present invention proposes a method for detecting workers' syncope based on pose, which mainly improves the pose detection model and algorithm logic analysis.

[0051] The following specifically illustrates the implementation details of the above method through embodiments. The following content is only the implementation details provided for convenient understanding and is not necessary for implementing this solution.

[0052] Embodiment 1

[0053] This embodiment provides a method for detecting workers' syncope based on pose, which mainly includes the following steps:

[0054] 1. Obtain the image information of the target area;

[0055] 2. Use a preset pose detection model to detect and track the pose of the person in the target area according to the obtained image information; wherein, the pose detection model is established based on an improved YOLO-POSE structure, and in the improved YOLO-POSE structure, multi-level feature maps introduce weights through BiFPN, and feature fusion at different scales is achieved through the weights;

[0056] 3. Count the pose key points of the target person and analyze the moving distance of these pose key points within a preset time period, and judge whether the pose key points are in a static state through the moving distance;

[0057] IV. Count the number of pose key points in the static state, and determine whether the target person is in a fainting state according to whether the number of pose key points in the static state exceeds a preset number threshold;

[0058] V. When the target person is in a fainting state, determine whether to send an alarm message.

[0059] In the above steps, target detection, pose estimation, and tracking mainly involve the pose detection model and the algorithm processing process.

[0060] Regarding the pose detection model:

[0061] The YOLO series of algorithms is a model that combines detection speed and accuracy and has wide applications in the industrial field. In this embodiment, an improved YOLOv8 algorithm is used to construct a pose detection model: a BiFPN (Bidirectional Feature Pyramid Network) structure is introduced between the backbone network and the detection head in the YOLOv8-POSE pose detection model. This improvement can significantly enhance the ability of the pose detection model in multi-scale feature fusion, especially being able to learn the context features between small and large targets. The pose detection model established in this way can more effectively integrate feature information from different resolutions. Especially for the human pose key points with partial position overlap, its detection effect is better. In addition, the weighted feature fusion mechanism of BiFPN allows the model to adaptively adjust the importance of different features, which further optimizes the accuracy of key point positioning. The pose detection model established by this improved YOLOv8 algorithm not only improves the detection efficiency but also can provide more robust pose estimation results in complex scenarios, especially performing more excellently in the application scenario of actual worker fainting detection.

[0062] Regarding the algorithm processing process: Security cameras can be installed in the workshop, and a vision server is established to process the real-time video stream and process the video stream in real time according to the preset algorithm process. First, use the detection algorithm to frame the people in the target area of each frame of the image and the pre-specified pose key points of the people; secondly, use the tracking algorithm to match and assign ids to the people appearing in consecutive frames of the image; then, count the pose change situation of the continuously appearing ids within a period of time and infer whether they are in a fainting state (such as the fainting state of falling to the ground); finally, according to whether there is a situation of a person fainting in the current frame, count the total number of frames with fainting. If the number of frames with consecutive people fainting in the specified area meets the preset threshold, an alarm event is pushed ( Figure 1 ).

[0063] In addition, in this embodiment, the judgment basis and related parameters for the falling process of a person can be specifically set as follows: for each detected target person, a person trajectory ID is assigned, the positions of the human key points of this ID within a preset time period (for example, thirty consecutive frames) are recorded, the moving distances of each key point in the first frame image and the last frame image are calculated, as well as the number of key points whose moving distances are greater than a preset distance threshold, so as to judge whether the person is in a fainting state.

[0064] Embodiment 2

[0065] The following combines a specific example to further illustrate the technical solution of the present invention and its beneficial effects.

[0066] As Figure 1 shown, a pose-based worker fainting detection method provided in this embodiment mainly includes the following steps.

[0067] I. Collecting workshop site images

[0068] As Figure 2 shown, in this embodiment, a security camera installed above the workshop is used to collect images. The camera view at this position can relatively completely cover most areas of the workshop when there is no lifting ladder blocking, the field of view is relatively open, and security cameras are provided on both sides of the workshop.

[0069] II. Personnel pose detection and tracking

[0070] In this step, an improved YOLOv8 algorithm is mainly used to implement personnel pose detection and tracking, and the BoT-SORT algorithm is combined to implement the detection box for tracking. The specific algorithm processing flow is as Figure 3 shown.

[0071] As mentioned above, the pre-trained pose model provided by the YOLOv8 algorithm has unsatisfactory effects in the actual production process, with many missed detections. In addition, in a complex production environment, when workers have target overlaps and / or targets are blocked, the detection effect is not good, there are missed detections and false alarms of mistaking columns and paint buckets for workers. Therefore, the present invention has improved the original YOLOv8 algorithm. As Figure 4 shown, the improvement of this algorithm is mainly in the enhanced feature extraction network, changing the original feature fusion method to introducing the BiFPN feature fusion method on feature maps of multiple levels. The pose detection model established in this way can perform feature fusion at different scales, improve the detection accuracy of small objects, and at the same time reduce information loss between different scales. The use of BiFPN helps to optimize the fusion process of feature maps, and through weighted fusion, the feature maps can be made more balanced. The new model structure diagram is as Figure 5 shown:

[0072] Among them, the structure of the feature fusion method based on BiFPN is as follows:

[0073] The specific implementation steps are as follows:

[0074] Let the input be x = [x 0 , x 1 , where X 0 and x 1 are two parts of the input tensor, that is, two feature maps to be fused.

[0075] Weight normalization: Given the weight vector w = [w 1 , w 2 , w 3 , calculate the effective weights through the normalization process as follows.

[0076]

[0077] Among them, ∈ is a small constant used to prevent division by zero errors.

[0078] Weighted fusion: Use the normalized weights to weight x 0 and x 1 .

[0079] x' 0 = weight 0 ·x 0

[0080] x' 1 = weight 1 ·x 1

[0081] Finally, concatenate the weighted x' 0 and x' 1 along the specified dimension d to obtain the output tensor as follows:

[0082] output = concat(x' 0 , x' 1 , dim = d)

[0083] The feature layer output in this way is a fusion feature layer that combines shallow features and deep features. Among them, the lower-level features are more detailed, but may not contribute highly in large object detection, while the higher-level features may contain more semantic information and are more important in large object detection. After introducing weights in this way, each layer of feature map can automatically adjust its contribution according to its importance in the current task, thereby improving the detection accuracy in cases such as small object detection, large object detection, and complex background processing.

[0084] According to Figure 6It can be intuitively seen that the detection accuracy of the human category has been significantly improved, especially for occluded and half-exposed human bodies. The improved model has better detection results.

[0085] Through Figure 7 and Figure 8 It can be intuitively seen that some workers wearing work clothes with partial occlusion have higher confidence. Moreover, even if only the head appears in the field of view, it can be judged with relatively high confidence, which is convenient for setting the strict and loose alarm modes through the confidence level during the subsequent alarm logic processing. In addition, the detection quantity of the 17 pose key points of the human body is also more accurate, alleviating the problem of difficult to determine the threshold of the number of static key points.

[0086] After achieving high-precision object detection and pose estimation, the tracking algorithm uses the tracking algorithm to predict the position of the detection box in the next frame, match it with the actual detection result in the next frame, and assign an id to the matched detection box, so as to realize the functions of tracking and counting. When a track id does not match the detection box within 30 frames, a new track id will be assigned to the current detection box.

[0087] Statistical distance of the seventeen pose key points of personnel moving within a certain period of time

[0088] After assigning an id to each detection box, the algorithm will record the positions of the seventeen key points of each id in the consecutive frames in the image. If there are key points that are not drawn due to limited perspective of the personnel in the image, the undrawn key points will be assigned [0,0]. The seventeen key points include the nose and the eyes, ears, shoulders, elbows, wrists, hips, knees and ankles on both sides. The algorithm will compare the Euclidean distances of these seventeen key points on the image before and after thirty frames, as shown in the following formula:

[0089]

[0090] Where P i,x and P i,y are the horizontal and vertical coordinates of the i-th key point in history, and C j,x and C j,y are the horizontal and vertical coordinates of the current j-th key point. The algorithm maintains a frame number queue with a maximum length of thirty to ensure that the track length of each id stored is thirty frames. Among them, i is set to the position of the first point in the algorithm, j is set to the position of the last point in the queue, and x and y respectively correspond to the horizontal and vertical coordinates of this point on the image.

[0091] Then compare the moving distance of each key point with the threshold. If it is lower than the threshold, this key point is regarded as not moving during this period of time. If it is higher than the threshold, it is considered that this part is moving. The formula is as follows:

[0092]

[0093] Among them, N represents the number of key points, and 1(d i,j <T) is an indicator function that takes the value of 1 when d i,j is less than the threshold T, and 0 otherwise. In this way, the number of key points that have not moved is calculated.

[0094] Judge whether the person is fainting and whether they are in the specified area according to the number of stationary key points

[0095] According to whether the key points move, count the key points that have hardly moved, so as to further judge the state of the person. If the following conditions are met, that is, the number of key points with a movement distance less than the threshold T exceeds the preset P threshold, it is considered that a person fainting is detected:

[0096] if stop_point_count > P

[0097] Among them, P is the number threshold of stationary key points.

[0098] When the number of stationary key points exceeds the threshold, it is considered that the person is fainting. To prevent false alarms in non-operation areas, a specific detection area is drawn on the picture. Only when a person fainting appears in this area, the current frame is regarded as an abnormal frame.

[0099] Judge whether to send an alarm according to the number of frames in which a person faints and de-duplicate the id

[0100] To prevent repeated alarms for one track id and sudden appearance of several abnormal frames, a logic for the threshold of the number of fainting frames and de-duplication of fainting person ids is designed. When the id of the fainting person appears for the first time and the number of consecutive abnormal frames is greater than the threshold, it is considered that the person faints within a period. The algorithm will output a warning text in the center of the detection box and send an alarm message to the platform, including information such as the event area, alarm time, and event.

[0101] Apply the trained model to the actual production environment for real-time inference, and the detection results are as Figure 9 shown. All thresholds are set as adjustable parameters, and the strict and loose modes of the trigger conditions can be adjusted according to the actual situation in the workshop. The model has achieved high accuracy and shown good robustness in the pose detection and tracking tasks. It can detect people with different perspectives and reach a high confidence level, and the pose is relatively more accurate.

[0102] Embodiment III

[0103] Another embodiment of the present invention relates to a pose-based worker fainting detection device, which includes:

[0104] An image information acquisition module for acquiring image information of the target area;

[0105] The pose detection and tracking module is used to detect and track the pose of a person in a target area according to the acquired image information by using a preset pose detection model; wherein, the pose detection model is established based on an improved YOLO-POSE structure, and in the improved YOLO-POSE structure, multi-level feature maps introduce weights through BiFPN, and feature fusion at different scales is achieved through the weights;

[0106] The static state judgment module is used to analyze the moving distance of the pose key points of the target person within a preset time period, and judge whether the pose key points are in a static state based on the moving distance;

[0107] The syncope state judgment module is used to judge whether the target person is in a syncope state according to the number of pose key points in a static state.

[0108] Embodiment 4

[0109] Another embodiment of the present invention relates to an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the pose-based worker syncope detection method in the above embodiments.

[0110] Among them, the memory and the processor are connected in a bus manner. The bus can include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits together, which are well known in the art, so they will not be further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be an element or multiple elements, such as multiple receivers and transmitters, and provides a unit for communicating with various other devices on the transmission medium. The data processed by the processor is transmitted on the wireless medium through the antenna. Further, the antenna also receives data and transmits the data to the processor.

[0111] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interface, voltage regulation, power management, and other control functions. The memory can be used to store the data used by the processor when executing operations.

[0112] Embodiment 5

[0113] Another embodiment of the present invention relates to a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the above-described embodiment of the pose-based worker syncope detection method.

[0114] That is, those skilled in the art can understand that all or part of the steps in implementing the above-described embodiment methods can be completed by instructing relevant hardware through a program. This program is stored in a storage medium and includes several instructions to enable a device (which can be a single-chip microcomputer, a chip, etc.) or a processor to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs that can store program codes.

[0115] Embodiment Six

[0116] Another embodiment of the present invention relates to a computer program product including a computer program. When the computer program is executed by a processor, it implements the steps of the above-described embodiment of the pose-based worker syncope detection method.

[0117] Those of ordinary skill in the art can understand that the above-described embodiments are specific embodiments for implementing the present invention. In actual applications, various changes can be made in form and details without departing from the spirit and scope of the present invention.

Claims

1. A method for detecting worker syncope based on posture, characterized in that: include: Acquire image information of the target area; Using a preset posture detection model, the posture of the person in the target area is detected and tracked according to the acquired image information; wherein the posture detection model is established based on an improved YOLO-POSE structure, in which multi-level feature maps in the improved YOLO-POSE structure introduce weights through BiFPN, and feature fusion at different scales is achieved through the weights; Analyze the moving distance of the target person's posture key points within a preset time period, and determine whether the posture key points are in a stationary state based on the moving distance; Whether the target person is in a fainting state is determined based on the number of posture key points in a static state.

2. The method for detecting worker syncope based on posture according to claim 1, characterized in that: The weight is used to characterize the importance of the feature in affecting the detection result.

3. The method for detecting worker syncope based on posture according to claim 1, characterized in that: The analysis of the moving distance of the target person's posture key points within a preset time period includes: calculating the Euclidean distance of each posture key point on the image before and after a number of frames according to the following formula: Among them, P i,x and P i,y is the horizontal and vertical coordinates of the key point of the pose in the i-th frame of history, C j,x and C j,y It is the horizontal and vertical coordinates of the key point of the current j-th frame pose.

4. The method for detecting worker syncope based on posture according to claim 3, characterized in that: The determining whether the posture key point is in a stationary state by the moving distance includes: Compare the moving distance of each pose key point with its corresponding distance threshold; If the moving distance is lower than the corresponding distance threshold, it is determined that the posture key point is in a stationary state; If the moving distance is greater than or equal to the corresponding distance threshold, it is determined that the posture key point is in a non-stationary state.

5. The method for detecting worker syncope based on posture according to claim 1, characterized in that: The step of judging whether the target person is in a fainting state according to the number of posture key points in a static state includes: When the number of posture key points in a static state exceeds a preset threshold, it is determined that the target person is in a fainting state.

6. The method for detecting worker syncope based on posture according to claim 1, characterized in that: The method further includes: when the target person is in a syncope state, determining whether to send an alarm message.

7. The method for detecting worker syncope based on posture according to claim 6, characterized in that: When the target person is in a syncope state, determining whether to send an alarm message includes: Count the number of image frames in which the target person is in a fainting state; When the target person faints for the first time and the number of image frames in the fainting state exceeds the preset frame number threshold, an alarm message is sent.

8. A posture-based worker syncope detection device, characterized in that: include: An image information acquisition module is used to obtain image information of a target area; A posture detection and tracking module is used to detect and track the posture of people in the target area according to the acquired image information using a preset posture detection model; wherein the posture detection model is established based on an improved YOLO-POSE structure, in which multi-level feature maps in the improved YOLO-POSE structure introduce weights through BiFPN, and feature fusion at different scales is achieved through the weights; A stationary state judgment module is used to analyze the moving distance of the target person's posture key points within a preset time period, and judge whether the posture key points are in a stationary state according to the moving distance; The syncope state judgment module is used to judge whether the target person is in a syncope state according to the number of posture key points in a static state.

9. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the posture-based worker syncope detection method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for detecting worker syncope based on posture according to any one of claims 1 to 7 is implemented.

Images