Attitude detection method and device, storage medium and electronic equipment

Abstract

The invention provides a posture detection method and device, a storage medium and electronic equipment, and the method is applied to the technical field of computers, and comprises the steps: employing a camera to collect a posture image of a target object, carrying out the feature extraction of the posture image, obtaining the posture feature of the target object, and obtaining a posture feature of the target object; and obtaining a target attitude category label matched with the attitude feature, and determining and outputting a target attitude prompt voice based on the target attitude category label. The method can collect the image to analyze the sitting posture of the target object and perform voice prompt, improves the accuracy of abnormal sitting posture recognition, enables the target object to adjust the sitting posture in time, guarantees the normal development of bones, and prevents the health problem caused by the abnormal sitting posture.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and more specifically, to a posture detection method, device, storage medium, and electronic device in the field of computer technology. Background Art

[0002] Sitting posture has an important impact on children's bone development. Bad sitting posture can lead to a series of bone development problems in children. Children in the learning stage need to sit for a long time, whether in school or at home. In order to enable children to maintain a correct sitting posture, schools and parents in the existing technology will choose to purchase chairs that help children correct their sitting posture. However, due to reasons such as sitting for too long, being too involved in learning, and lack of concentration, it is difficult for children to rely on self-control to maintain a correct sitting posture. Therefore, a method is needed to detect and prompt abnormal sitting posture. Summary of the Invention

[0003] The embodiments of the present application provide a posture detection method, device, storage medium and electronic device. The method can analyze the sitting posture of the target object by collecting images and provide voice prompts, thereby improving the accuracy of abnormal sitting posture recognition and enabling the target object to adjust the sitting posture in time, thereby ensuring normal bone development and preventing health problems caused by abnormal sitting posture.

[0004] In a first aspect, an embodiment of the present application provides a posture detection method, the method comprising:

[0005] A camera is used to collect the posture image of the target object;

[0006] Performing feature extraction processing on the posture image to obtain posture features of the target object;

[0007] Obtaining a target posture category label that matches the posture feature;

[0008] A target posture prompt voice is determined and output based on the target posture category label.

[0009] Through the above technical solution, the target object's sitting posture can be analyzed based on the collected image and voice prompts can be given, which improves the accuracy of abnormal sitting posture recognition and enables the target object to adjust its sitting posture in time, thereby ensuring normal bone development and preventing health problems caused by abnormal sitting posture.

[0010] In combination with the first aspect, in some possible implementations, the posture feature includes at least one of a head posture feature, a shoulder posture feature, a back posture feature, a leg posture feature, and an overall posture feature.

[0011] Through the above technical solution, the posture features of different body parts of the target object can be extracted. Through multi-dimensional feature extraction, the misjudgment that may be caused by a single feature can be reduced. At the same time, the individual characteristics of different objects are met, and the accuracy of the sitting posture detection results is improved.

[0012] In conjunction with the first aspect, in some possible implementations, before obtaining the target posture category label that matches the posture feature, the method further includes:

[0013] Acquiring pressure sensing data of the target object using a pressure sensor;

[0014] The obtaining of a target posture category label that matches the posture feature includes:

[0015] Obtain a target posture category label that matches the pressure sensing data and the posture feature.

[0016] Through the above technical solution, the sitting state of the target object can be determined by combining pressure sensing data and posture features. The use of multimodal data fusion improves the comprehensiveness of sitting posture-related features and further improves the accuracy of posture detection.

[0017] In combination with the first aspect, in some possible implementations, the pressure sensing data includes at least one of front end pressure data, rear end pressure data, left end pressure data and right end pressure data on the seat surface of the target object seat.

[0018] Through the above technical solution, different pressure data can be obtained according to pressure sensors in different positions, the pressure conditions of various parts of the target object's body can be detected, and the sitting posture of the target object can be evaluated in multiple dimensions, further reducing misjudgment.

[0019] In conjunction with the first aspect, in some possible implementations, performing feature extraction processing on the posture image to obtain the posture features of the target object includes:

[0020] inputting the posture image and the pressure sensing data into a posture recognition model;

[0021] Acquiring a posture feature of the object based on the posture recognition model;

[0022] The acquiring of a target posture category label that matches the pressure sensing data and the posture feature includes:

[0023] Obtain the target posture category label output by the posture recognition model.

[0024] Through the above technical solution, the model can participate in the process of feature extraction and data analysis, which improves the efficiency of multimodal data processing and the accuracy of posture category label recognition.

[0025] In conjunction with the first aspect, in some possible implementations, before using a camera to capture a posture image of the target object, the method further includes:

[0026] Creating an initial posture recognition model, obtaining a sample posture image, a sample posture category label corresponding to the sample posture image, and sample pressure sensing data corresponding to the sample posture image;

[0027] Inputting the sample posture image and the sample pressure sensing data into the initial posture recognition model to obtain a training posture category label output by the initial posture recognition model;

[0028] Based on the sample posture category labels and the training posture category labels, parameter adjustment processing is performed on the initial posture recognition type until model training is completed to obtain a posture recognition model.

[0029] Through the above technical solution, model training can be performed based on sample posture images and sample pressure sensing data, thereby improving the accuracy and reliability of the posture recognition model, making the posture recognition model adaptable to different application environments and improving the generalization ability of the model.

[0030] In conjunction with the first aspect, in some possible implementations, determining and outputting a target posture prompt voice based on the target posture category label includes:

[0031] Determining a target sitting posture state corresponding to the target posture category label, and obtaining a sitting posture state correction suggestion corresponding to the target sitting posture state;

[0032] Output a target posture prompt voice based on the sitting posture correction suggestion.

[0033] Through the above technical solution, a target posture prompt voice containing sitting posture correction suggestions can be output, giving the target object specific correction suggestions, which is convenient for the target object to understand and execute, helping the target object to adjust the sitting posture in time and prevent health problems.

[0034] In a second aspect, an embodiment of the present application provides a posture detection device, the device comprising:

[0035] An image acquisition unit, configured to acquire a posture image of a target object using a camera;

[0036] a posture feature extraction unit, configured to perform feature extraction processing on the posture image to obtain posture features of the target object;

[0037] A posture category acquisition unit, configured to acquire a target posture category label that matches the posture feature;

[0038] A voice prompt unit is used to determine and output a target posture prompt voice based on the target posture category label.

[0039] In a third aspect, an embodiment of the present application provides a computer storage medium, wherein the computer storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor and executing the above-mentioned method steps.

[0040] In a fourth aspect, an embodiment of the present application provides an electronic device, which may include: a processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the above-mentioned method steps.

[0041] In one or more embodiments of the present application, a camera is used to capture a posture image of a target subject, and feature extraction processing is performed on the posture image to obtain posture features of the target subject, obtain a target posture category label that matches the posture features, and determine and output a target posture prompt based on the target posture category label. By capturing images, analyzing the target subject's sitting posture, and providing a voice prompt, the accuracy of abnormal sitting posture recognition is improved, allowing the target subject to adjust their sitting posture in a timely manner, thereby ensuring normal skeletal development and preventing health problems caused by abnormal sitting posture. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0043] Figure 1 This is a schematic diagram of a scene for capturing an image of a target object provided by an embodiment of the present application;

[0044] Figure 2 1 is a flow chart of a posture detection method provided in an embodiment of the present application;

[0045] Figure 3 This is a flow chart of a model training process provided by an embodiment of the present application;

[0046] Figure 4 This is a flow chart of a prompt voice output provided by an embodiment of the present application;

[0047] Figure 5 Schematic diagram of a posture detection device provided in an embodiment of the present application;

[0048] Figure 6 Schematic diagram of a posture detection device provided in an embodiment of the present application;

[0049] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0051] In daily work and life, people may need to sit for long periods of time, and a standard sitting posture is very important for human health, especially for children in the skeletal development stage. For example, a child's spine has a high degree of plasticity during growth and development. Poor sitting postures, such as hunching over or slumping, can cause changes in the normal physiological curvature of the spine and may even cause scoliosis, affecting height development and other problems. Long-term spinal deformation not only affects a child's appearance but also has a negative impact on bodily functions such as cardiopulmonary function. Parents and teachers should teach children to maintain a normal sitting posture. However, in scenarios such as learning where children need to sit for long periods of time, they are still prone to various poor sitting postures due to their immature physical development, easily distracted attention, and lack of knowledge of correct sitting posture. The present application provides an embodiment of a posture detection device that can use a camera to capture an image of a target object and identify the target object's sitting posture, thereby providing a prompt when the target object adopts an unhealthy sitting posture. The posture detection device can be used in home scenarios or school teaching scenarios, and the target object can be a child or student. The posture detection method provided in the embodiment of the present application can be implemented by a computer program and can be run on a posture detection device based on the von Neumann architecture. This computer program can be integrated into an application or run as an independent tool application. The posture detection device can call and control the camera. In addition to capturing the sitting image of the target object, the camera can also monitor the target object to ensure the safety of the target object in home scenes or school teaching scenes.

[0052] Please also see Figure 1, provides a scene diagram of target object image acquisition for the embodiment of the present application. The camera can be installed by a guardian, school staff, etc. The camera's viewing angle can cover the target object's desk and seat range, making it easy to capture the target object's sitting posture. For example, the camera's viewing angle can include various parts of the target object's body, such as shoulders, back, waist and legs, etc., and can also include the relative position of the desk and seat. When the target object sits on the seat, the posture detection device can control the camera to capture the target object's posture image, such as Figure 1 The sitting posture image A and the sitting posture image B in the figure may be posture images captured successively by the camera. In posture image A, the target object sits upright, and in posture image B, the target object sits hunched over. The posture detection device may perform posture detection processing on the sitting posture image after acquiring the sitting posture image, thereby determining that the sitting posture of the target object in sitting posture image A is correct. At this time, the posture detection device may not give a prompt to avoid disturbing the normal work and study of the target object. When it is determined that the target object in sitting posture image B has a bad sitting posture, a prompt voice may be output to remind the target object to correct the current bad sitting posture, thereby preventing the target object from having health problems caused by long-term bad sitting posture.

[0053] The posture detection method provided in this application is described in detail below with reference to specific embodiments.

[0054] See Figure 2 , which is a flow chart of a posture detection method provided in the embodiment of the present application. Figure 2 As shown, the method of the embodiment of the present application may include the following steps S101-S104.

[0055] S101, using a camera to collect a posture image of a target object.

[0056] Specifically, when the target object enters the camera's field of view and sits on the seat, the posture detection device can control the camera to capture the target object's posture image. The posture image is used to show the target object's sitting posture, and may include, for example, various parts of the target object's body, such as shoulders, back, waist, and legs. For the purpose of prompting the target object to maintain a correct sitting posture, the posture detection device can control the camera to continuously capture the target object's posture image. For example, the target object's posture image can be captured at a preset acquisition frequency, wherein the preset acquisition frequency may be related to the hardware conditions of the camera, or may be related to the computing power of the posture detection device. It may be an initial setting of the posture detection device, or may be set by the target object's guardian or relevant staff.

[0057] S102: Perform feature extraction processing on the posture image to obtain posture features of the target object.

[0058] Specifically, the posture image may include various parts of the target object's body, so the posture detection device can perform feature extraction processing on the posture image, analyze the status of various parts of the target object's body, and thus obtain the posture characteristics of the target object. The posture characteristics can be feature vectors used to characterize the sitting posture of the target object in the posture image. For example, the posture characteristics can be a multi-dimensional numerical vector, each dimension can represent a specific feature of the target object's sitting posture, and the specific features can include information such as the position and angle of various parts of the target object's body, thereby comprehensively and quantitatively describing the target object's sitting posture.

[0059] S103: Obtain a target posture category label that matches the posture feature.

[0060] Specifically, the posture detection device can classify the possible sitting postures of the human body to obtain multiple posture category labels. The posture category label is used to represent a sitting posture state and can be a serial number or text. It can be the initial setting of the posture detection device or set by the guardian or relevant staff of the target subject. The posture detection device can determine the corresponding sitting posture state based on the posture characteristics, and thus determine a matching target posture category label. The target posture category label is the posture category label determined by the posture detection device that matches the sitting posture state in the posture image.

[0061] S104: Determine and output a target posture prompt voice based on the target posture category label.

[0062] Specifically, the posture category labels may include posture category labels corresponding to normal sitting posture states, and posture category labels corresponding to poor sitting posture states. If the target posture category label is a posture category label corresponding to a poor sitting posture state, a target posture prompt language may be output. The target posture prompt voice is used to prompt the target object to adjust the current sitting posture state. For example, it may be "Sit upright, otherwise it will affect your vision and physical development."

[0063] Optionally, since different posture category labels can correspond to different sitting postures, in order to help the target object adjust from different sitting postures to a normal sitting posture, different posture prompt voices can be set for different posture category labels. The posture prompt voices can also include sitting posture correction suggestions to help the target object adjust the sitting posture corresponding to the posture category label. For example, if the posture category label is "body leaning forward", the corresponding sitting posture can be that the body is leaning forward, the upper body is almost lying on the table, the head is very close to the table, and the back is bent. The corresponding posture prompt voice can be "The current sitting posture is leaning forward, please lean back and keep an upright sitting posture." Conversely, if the posture category label is "body leaning back", the corresponding sitting posture can be that the body is leaning back, sitting at the back of the chair, with a large gap between the back and the back of the chair, and sometimes even the head is leaning on the back of the chair. The corresponding posture prompt voice can be "The current sitting posture is leaning back, please lean forward and keep an upright sitting posture." Therefore, the posture detection device can determine the corresponding target posture prompt voice according to the target posture category label, and then output the target posture prompt voice, while prompting the target object that there is a problem with the current sitting posture state, and at the same time, making correction suggestions for the sitting posture state.

[0064] In this embodiment, a camera is used to capture a posture image of a target subject. Feature extraction is performed on the posture image to obtain the target subject's posture features, a target posture category label that matches the posture features is obtained, and a target posture prompt voice is determined and output based on the target posture category label. By capturing images and analyzing the target subject's sitting posture and providing voice prompts, the accuracy of abnormal sitting posture recognition is improved, allowing the target subject to adjust their sitting posture in a timely manner, thereby ensuring normal skeletal development and preventing health problems caused by abnormal sitting posture.

[0065] The camera's field of view can include all parts of the target object's body, so that the posture detection device can comprehensively judge the target object's sitting posture based on the postures of various parts of the target object's body, thereby improving the accuracy of the sitting posture judgment. Therefore, the posture feature device can perform feature extraction processing on each part.

[0066] In one or more embodiments of the present application, the posture feature includes at least one of a head posture feature, a shoulder posture feature, a back posture feature, a leg posture feature, and an overall posture feature.

[0067] Specifically, after acquiring the posture image, the posture detection device can determine the head, shoulders, back, legs and / or the entire body of the target object in the posture image, and then extract the head posture features, shoulder posture features, back posture features, leg posture features and overall posture features respectively, wherein the head posture features may include information such as the head position, the relative angle between the head and the body, the relative angle between the head and the shoulders and the neck, etc., the shoulder posture features may include information such as the shoulder position, the shoulder horizontality, the relative angle between the shoulder and the back and the waist, etc., the back posture features may include information such as the back contour morphology, the physiological curvature of the spine, the contact area between the back and the seat back, etc., the leg posture features may include information such as the waist position, whether the waist is in contact with the seat back, etc., the leg posture features may include information such as the leg position, leg posture, whether the legs are placed on the ground, etc., and the overall posture features may include information such as the position of the center of gravity of the body, the symmetry of the body, and the relative position of the body and the desk. The posture features of different parts complement each other and provide a comprehensive and comprehensive detection result. For example, forward head tilt may occur at the same time as shoulder extension. Combining these posture features can more accurately determine whether the sitting posture is correct. In addition, the body structure and habits of different subjects may be different. Through multi-dimensional posture feature extraction, it can better adapt to the characteristics of different subjects.

[0068] Optionally, due to issues such as the limited camera viewing angle or obstructions such as tables and chairs, the captured posture image may not simultaneously include the target subject's head, shoulders, back, and legs. Therefore, the posture detection device may only obtain posture features corresponding to the existing parts. The posture detection device may perform body part detection processing on the posture image to determine which parts are present in the posture image and the locations of each part, and then extract corresponding posture features based on the locations of each part.

[0069] In an embodiment of the present application, posture features of different body parts of the target object are extracted. Through multi-dimensional feature extraction, misjudgment that may be caused by a single feature is reduced. At the same time, the individual characteristics of different objects are met, and the accuracy of the sitting posture detection results is improved.

[0070] Detecting sitting posture based solely on sitting posture images may be overly simplistic. The posture detection device can also combine posture characteristics with the pressure exerted by the subject on the seat. This multimodal data fusion can be used to determine the subject's sitting posture, thereby improving posture detection accuracy. The subject's seat can be equipped with a pressure sensor, and the posture detection device can use this pressure sensor to detect the pressure.

[0071] In one or more embodiments of the present application, before step S103, the following steps are further included:

[0072] A pressure sensor is used to obtain pressure sensing data of the target object.

[0073] Specifically, there may be a pressure sensor on the seat of the target object. For example, the pressure sensor may be installed on the seat surface to detect the pressure when the target object sits on the seat. When the target object sits on the seat, the posture detection device may use a camera to capture the posture image of the target object, and may also use a pressure sensor to obtain the pressure sensing data of the target object at the same time.

[0074] Optionally, the pressure sensor can also collect pressure sensing data according to a preset collection frequency, so the pressure sensing data collected by the pressure sensor corresponds one-to-one to the posture image and points to the same sitting posture state, thereby ensuring the uniformity of posture characteristics and pressure sensing data.

[0075] The step S103 may include the following steps:

[0076] Obtain target pose category labels that match pressure sensing data and pose features.

[0077] Specifically, the posture detection device can combine the pressure sensing data and the posture features to determine the sitting posture state of the target object in the posture image, thereby obtaining a matching target posture category label.

[0078] Optionally, the posture detection device can perform feature extraction processing based on the target object's pressure sensing data to obtain pressure characteristics corresponding to the target object. The pressure characteristics can be used to characterize the target object's pressure distribution relative to the seat. It can be understood that the pressure distribution can determine whether various body parts of the target object are supported by the seat, thereby indirectly reflecting the target object's sitting posture. Furthermore, the posture detection device can perform feature fusion processing on the posture characteristics and pressure characteristics to obtain a comprehensive feature vector, and then determine a matching target posture category label based on the comprehensive feature vector.

[0079] In the embodiment of the present application, the sitting state of the target object is determined by combining pressure sensing data and posture features, and multimodal data fusion is used to improve the comprehensiveness of sitting posture-related features and further improve the accuracy of posture detection.

[0080] In order to obtain a more comprehensive pressure distribution and further understand the support conditions of the target object's various body parts by the seat, multiple pressure sensors can be installed at different positions on the target object's seat. For example, pressure sensors can be installed at different positions such as the front, back, left, and right of the seat surface to obtain multiple pressure data, and the pressure sensing data can be obtained by integrating the multiple pressure data.

[0081] In one or more embodiments of the present application, the pressure sensing data includes at least one of front end pressure data, rear end pressure data, left end pressure data and right end pressure data on the seat surface of the target object seat.

[0082] Specifically, pressure sensors can be installed at different locations on the chair surface, such as the front, back, left, and right. Based on these pressure sensors, the posture detection device can obtain front-end pressure data, back-end pressure data, left-end pressure data, and / or right-end pressure data. For example, when the difference between the front-end pressure data and the back-end pressure data is positive and greater than a preset pressure value, it can be determined that the target subject is likely to be leaning forward, and the judgment can be further combined with posture characteristics. Conversely, when the difference between the back-end pressure data and the front-end pressure data is positive and greater than a preset pressure value, it can be determined that the target subject is likely to be leaning forward, and the judgment can be further combined with posture characteristics. Similarly, when the difference between the left-end pressure data and the right-end pressure data is positive and greater than a preset pressure value, it can be determined that the target subject is likely to be leaning to the left. When the difference between the right-end pressure data and the left-end pressure data is positive and greater than a preset pressure value, it can be determined that the target subject is likely to be leaning to the right.

[0083] Optionally, in addition to arranging pressure sensors on the seat surface, pressure sensors can also be installed on the seat back, armrests and other positions to comprehensively collect pressure data generated by the contact between various parts of the target object's body and the seat, and further accurately analyze the support conditions of various parts of the body. For example, pressure sensors can be arranged at the upper, middle and lower positions of the seat back, so that the pressure sensing data can also include upper back pressure data, middle back pressure data and lower back pressure data. Pressure sensors can also be installed at the left and right armrests, so that the pressure sensing data can also include left arm pressure data and right arm pressure data. The posture detection device can combine all pressure data in the pressure sensing data to determine the pressure distribution and obtain pressure characteristics. Multi-dimensional pressure data can improve the accuracy of the pressure distribution.

[0084] In the embodiment of the present application, different pressure data can be obtained based on pressure sensors at different positions, and the pressure conditions of various parts of the target object's body can be detected, which helps to evaluate the target object's sitting posture in multiple dimensions and further reduces misjudgment.

[0085] In order to further improve the accuracy of sitting posture detection and improve computing efficiency, the posture detection device can train a convolutional neural network (CNN) to participate in the process of feature extraction and feature analysis. Since CNN is a deep learning model, it can be applied to fields such as image recognition and natural language processing. It can help extract and analyze the features of posture images and pressure sensing data.

[0086] In one or more embodiments of the present application, step S102 may include the following steps:

[0087] The posture image and pressure sensing data are input into a posture recognition model; and the posture features of the object are obtained based on the posture recognition model.

[0088] Specifically, the posture detection device can input the collected posture images and pressure sensing data into the posture recognition model. The posture recognition model can perform feature extraction processing on the posture images to obtain the posture features corresponding to the target object. Similarly, the posture recognition model can also perform feature extraction processing on the pressure sensing data to obtain the pressure features corresponding to the target object.

[0089] The step of obtaining a target posture category label that matches the pressure sensing data and the posture feature may include the following steps:

[0090] Get the target pose category label output by the pose recognition model.

[0091] Specifically, the posture recognition model can analyze the sitting posture state of the target object based on the posture image and pressure sensor data, and determine the matching target posture category label. After the posture detection device inputs the posture image and pressure sensor into the posture recognition model, it can obtain the target posture category label output by the posture recognition model.

[0092] In this embodiment, posture images and pressure sensor data are input into a posture recognition model. The posture features of the object are then extracted based on the posture recognition model, and the target posture category label is output by the posture recognition model. By incorporating the model into the feature extraction and data analysis process, the efficiency of multimodal data processing is improved, as is the accuracy of posture category label recognition.

[0093] See Figure 3 , provides a flow chart of model training for the embodiment of this application. Figure 3 As shown, in one or more embodiments of the present application, the following steps S201-S203 may be included before step S101.

[0094] S201 , creating an initial posture recognition model, obtaining a sample posture image, a sample posture category label corresponding to the sample posture image, and sample pressure sensing data corresponding to the sample posture image.

[0095] Specifically, an initial posture recognition model can be created, which has the preliminary ability to match posture category labels based on posture images and pressure sensor data. Sample posture images, sample posture category labels corresponding to the sample posture images, and sample pressure sensor data corresponding to the sample posture images can also be obtained. The sample posture images can be real posture images taken by relevant staff of the sample objects. The sample posture category labels can be labels annotated by relevant staff for the sample posture images. The sample pressure sensor data is the pressure sensor data corresponding to the sample posture images.

[0096] Optionally, the sample posture images may include posture images that include the head, shoulders, back, legs, and entire body of the sample object, or posture images that only include part of the sample object's body, for example, posture images that only include the head and back of the sample object, or posture images that only include the head and legs of the sample object, etc., so that the posture recognition model obtained after model training can accurately match the target posture category label according to the unobstructed body parts when obtaining posture images with occluded body parts of the target object, thereby improving the robustness of the posture recognition model.

[0097] Similarly, the sample pressure sensor data may include pressure sensor data that includes front-end pressure data, rear-end pressure data, left-end pressure data, and right-end pressure data. It may also include pressure sensor data that includes, in addition to seat surface pressure sensor data, seat back pressure sensor data such as upper back pressure data, middle back pressure data, and lower back pressure data, as well as armrest pressure sensor data such as left arm pressure data and right arm pressure data. It may also include pressure sensor data that includes only partial sensor data from the seat surface pressure sensor data, seat back pressure sensor data, and armrest pressure sensor data. Because different seats may come from different manufacturers or be of different models, the number and location of pressure sensors on different target seats may vary. For example, some seats may only have seat surface pressure sensors, while others may have both seat surface and armrest pressure sensors. This allows the posture recognition model to adapt to different seats and, therefore, to different personalized environments, improving the model's generalization capabilities.

[0098] S202 : Input the sample posture image and the sample pressure sensing data into an initial posture recognition model to obtain a training posture category label output by the initial posture recognition model.

[0099] Specifically, the sample posture image and sample pressure sensor data can be input into the initial posture recognition model, and the initial posture recognition model can perform feature extraction processing on the sample posture image and sample pressure sensor data to determine the training posture category label that matches the sample posture image and sample pressure sensor data. The training posture category label is the posture category label predicted by the initial posture recognition model based on the current posture detection capability.

[0100] Optionally, the feature extraction and data analysis process of the initial posture recognition model for the sample posture image and sample pressure sensing data can refer to the above-mentioned embodiment. For example, the initial posture recognition model can extract sample posture features from the sample posture image, extract sample pressure features from the sample pressure sensing data, and then fuse the sample posture features and the sample pressure features to obtain sample comprehensive features, and then match the training posture category labels based on the sample comprehensive features.

[0101] S203 , based on the sample posture category labels and the training posture category labels, perform parameter adjustment processing on the initial posture recognition type until the model training is completed to obtain a posture recognition model.

[0102] Specifically, the loss function of the initial posture recognition model can be calculated based on the sample posture category labels and the training posture category labels, and the parameters of the initial posture recognition model can be adjusted based on the loss function during the back-propagation training process until the initial posture recognition model completes the model training to obtain the posture recognition model.

[0103] Optionally, based on the sample posture category labels and the training posture category labels, the parameters of the initial posture recognition type can be adjusted until the training termination condition is met to obtain a posture recognition model. The training termination condition may include the loss function falling below a preset loss value and maintaining it for a preset time period. The preset loss value is used to determine whether the prediction accuracy of the initial posture recognition model has met the standard, and the preset time period is used to determine whether the stability of the initial posture recognition model has met the standard. The preset loss value and the preset time period may be initial settings of the posture detection device or may be set by relevant staff. For example, the preset loss value may be 0.05, and the preset time period may be 100 rounds of model training. The training termination condition may also include the model training rounds for the initial posture recognition model reaching a preset number of rounds. For example, the preset number of rounds may be 100. Using the training rounds as the training termination condition is very intuitive and easy to understand, and does not require complex calculations or judgment conditions. The number of rounds only needs to be set at the beginning of training, making the management of the training process more concise.

[0104] In an embodiment of the present application, an initial posture recognition model is created, sample posture images, sample posture category labels corresponding to the sample posture images, and sample pressure sensor data corresponding to the sample posture images are obtained. The sample posture images and sample pressure sensor data are input into the initial posture recognition model, and a training posture category label output by the initial posture recognition model is obtained. Based on the sample posture category label and the training posture category label, parameters of the initial posture recognition type are adjusted until model training is completed, thereby obtaining a posture recognition model. Model training using sample posture images and sample pressure sensor data improves the accuracy and reliability of the posture recognition model, making it adaptable to different application environments and enhancing the model's generalization capabilities.

[0105] See Figure 4 , provides a flow chart of prompting voice output for the embodiment of the present application. Figure 4 As shown, in one or more embodiments of the present application, step S104 may include the following steps S301-S302.

[0106] S301, determining a target sitting posture state corresponding to a target posture category label, and obtaining a sitting posture correction suggestion corresponding to the target sitting posture state.

[0107] Specifically, each target posture category label corresponds to a sitting posture state. The posture detection device can set sitting posture correction suggestions for different sitting posture states. The sitting posture correction suggestions are used to prompt the target object to correct the current bad sitting posture state to a normal sitting posture state. Specific sitting posture correction suggestions can be provided, rather than simple vague prompts such as "sit well" or "adjust posture", which can help the target object understand more clearly how to adjust the sitting posture. For example, specific suggestions such as "please put your feet flat on the ground" or "the current sitting posture is leaning back, please keep your body forward and sit upright" allow the target object to intuitively understand his or her sitting posture state, and the correction suggestions can also be more easily understood and executed by the target object. Therefore, the sitting posture correction state can determine the target sitting posture state corresponding to the target posture category label, and then obtain the sitting posture correction suggestion corresponding to the target sitting posture state, wherein the sitting posture correction suggestion can be pre-set by relevant staff.

[0108] Optionally, the posture detection device can use a large language model (LLM) to generate sitting posture correction suggestions. The target sitting posture state can be input into the large language model, and the large language model can be prompted to generate correction suggestions based on the target sitting posture state, thereby obtaining the sitting posture correction suggestions output by the large language model.

[0109] S302: Outputting a target posture prompt voice based on the sitting posture correction suggestion.

[0110] Specifically, a target posture prompt voice can be output based on the sitting posture correction suggestion, that is, the target posture prompt voice contains the sitting posture correction suggestion, so that the target object can adjust back to the normal sitting posture according to the sitting posture correction suggestion after hearing the target posture prompt voice.

[0111] Optionally, the posture detection device may include at least one preset tone. The preset tone may be set by the posture detection device or relevant staff based on an electronic tone on the Internet, or may be input and set by the guardian or teacher of the target object. The target object or the guardian or teacher of the target object may select and set the target preset tone from the preset tones. The posture detection device may output a target posture prompt voice based on the target preset tone and the sitting posture correction suggestion. The posture detection device may generate a target posture prompt voice containing a sitting posture correction suggestion based on the target preset tone based on the Text-to-Speech (TTS) technology. Since the target object may be a child, the target preset tone may be used to simulate the voice of their guardian or teacher. While prompting the target object to maintain a normal sitting posture, it may also soothe the target object's emotions and meet the personalized needs of different objects.

[0112] In this embodiment of the present application, a target sitting posture state corresponding to a target posture category label is determined, a posture correction suggestion corresponding to the target sitting posture state is obtained, and a target posture prompt voice message is output based on the posture correction suggestion. By outputting a target posture prompt voice message containing the posture correction suggestion, the target subject is given specific correction suggestions, which are easy for the target subject to understand and implement, helping the target subject to adjust their sitting posture in a timely manner and prevent health problems.

[0113] The following will be combined with the Figure 5 -Attached Figure 6 , the posture detection device provided in the embodiment of the present application is introduced in detail. Figure 5 -Attached Figure 6 The posture detection device in the present application is used to execute Figure 1-4 Figure 2 For the convenience of explanation, only the part related to the embodiment of the present application is shown. For the specific technical details not disclosed, please refer to the present application. Figures 1-4 The embodiment shown.

[0114] See Figure 5 , which shows a schematic diagram of the structure of a posture detection device provided by an exemplary embodiment of the present application. The posture detection device can be implemented as all or part of the device through software, hardware, or a combination of both. The device 1 includes an image acquisition unit 11, a posture feature extraction unit 12, a posture category acquisition unit 13, and a voice prompt unit 14.

[0115] An image acquisition unit 11 is used to acquire a posture image of a target object using a camera;

[0116] A posture feature extraction unit 12 is used to perform feature extraction processing on the posture image to obtain the posture features of the target object;

[0117] A posture category acquisition unit 13 is used to acquire a target posture category label that matches the posture feature;

[0118] The voice prompt unit 14 is configured to determine and output a target posture prompt voice based on the target posture category label.

[0119] In this embodiment, a camera is used to capture a posture image of the target subject. Feature extraction is performed on the posture image to obtain the target subject's posture features, a target posture category label matching the posture features is obtained, and a target posture prompt voice is determined and output based on the target posture category label. By analyzing the target subject's sitting posture through image capture and providing voice prompts, the accuracy of abnormal sitting posture recognition is improved, enabling the subject to adjust their sitting posture in a timely manner, thereby ensuring normal skeletal development and preventing health problems caused by abnormal sitting posture.

[0120] See Figure 6 , which shows a schematic diagram of the structure of a posture detection device provided by an exemplary embodiment of the present application. The posture detection device can be implemented as all or part of the device through software, hardware, or a combination of both. The device 1 includes a model training unit 15, an image acquisition unit 11, a posture feature extraction unit 12, a pressure data acquisition unit 16, a posture category acquisition unit 13, and a voice prompt unit 14.

[0121] The model training unit 15 is used to create an initial posture recognition model, obtain sample posture images, sample posture category labels corresponding to the sample posture images, and sample pressure sensing data corresponding to the sample posture images;

[0122] Inputting the sample posture image and the sample pressure sensing data into the initial posture recognition model to obtain a training posture category label output by the initial posture recognition model;

[0123] Based on the sample posture category labels and the training posture category labels, parameter adjustment processing is performed on the initial posture recognition type until model training is completed to obtain a posture recognition model.

[0124] An image acquisition unit 11 is used to acquire a posture image of a target object using a camera;

[0125] A posture feature extraction unit 12 is used to perform feature extraction processing on the posture image to obtain the posture features of the target object;

[0126] Optionally, the posture feature includes at least one of a head posture feature, a shoulder posture feature, a back posture feature, a leg posture feature and an overall posture feature.

[0127] Optionally, the posture feature extraction unit 12 is specifically configured to input the posture image and the pressure sensing data into a posture recognition model;

[0128] Acquiring a posture feature of the object based on the posture recognition model;

[0129] A pressure data acquisition unit 16 is configured to acquire pressure sensing data of the target object using a pressure sensor;

[0130] Optionally, the pressure sensing data includes at least one of front end pressure data, rear end pressure data, left end pressure data and right end pressure data on the seat surface of the target object seat.

[0131] A posture category acquisition unit 13 is used to acquire a target posture category label that matches the posture feature;

[0132] Optionally, the posture category acquisition unit 13 is specifically configured to acquire a target posture category label that matches the pressure sensing data and the posture feature.

[0133] Optionally, the posture category acquisition unit 13 is specifically configured to acquire a target posture category label output by the posture recognition model.

[0134] The voice prompt unit 14 is configured to determine and output a target posture prompt voice based on the target posture category label.

[0135] Optionally, the voice prompt unit 14 is specifically configured to determine a target sitting posture state corresponding to the target posture category label, and obtain a sitting posture correction suggestion corresponding to the target sitting posture state;

[0136] Output a target posture prompt voice based on the sitting posture correction suggestion.

[0137] In this embodiment, an initial posture recognition model is created, sample posture images, sample posture category labels corresponding to the sample posture images, and sample pressure sensor data corresponding to the sample posture images are obtained. The sample posture images and sample pressure sensor data are input into the initial posture recognition model, and a training posture category label output by the initial posture recognition model is obtained. Based on the sample posture category label and the training posture category label, the parameters of the initial posture recognition type are adjusted until model training is completed, thereby obtaining a posture recognition model. Model training using sample posture images and sample pressure sensor data improves the accuracy and reliability of the posture recognition model, making the posture recognition model adaptable to different application environments and enhancing the model's generalization capabilities. A camera is used to capture posture images of a target object, and feature extraction is performed on the posture images to obtain posture features of the target object. These features are then extracted for different body parts of the target object. This multi-dimensional feature extraction reduces potential misjudgments caused by a single feature, while also addressing the individual characteristics of different subjects and improving the accuracy of sitting posture detection results. Pressure sensors are used to acquire pressure sensor data of the target object. Different pressure data can be obtained from pressure sensors located at different locations, allowing pressure conditions in different body parts of the target object to be detected. This facilitates multi-dimensional assessment of the target object's sitting posture, further reducing misjudgments. Posture images and pressure sensor data are input into a posture recognition model. The posture features of the subject are acquired based on the posture recognition model, and a target posture category label is output by the posture recognition model. The model's involvement in feature extraction and data analysis improves the efficiency of multimodal data processing and the accuracy of posture category label recognition. Combining pressure sensor data with posture features to determine the target subject's sitting posture, multimodal data fusion enhances the comprehensiveness of posture-related features, further improving posture detection accuracy. A target posture prompt is determined and output based on the target posture category label. By acquiring images and analyzing the target subject's sitting posture and providing voice prompts, the accuracy of abnormal sitting posture recognition is improved, enabling the subject to adjust their sitting posture in a timely manner, thereby ensuring normal skeletal development and preventing health problems caused by abnormal sitting posture. The target sitting posture corresponding to the target posture category label is determined, and posture correction suggestions corresponding to the target sitting posture are obtained. Based on the posture correction suggestions, a target posture prompt is output based on the posture correction suggestions. By outputting a target posture prompt voice containing the posture correction suggestions, the subject is provided with specific correction suggestions, making them easier to understand and implement.

[0138] It should be noted that the posture detection device provided in the above embodiment only uses the division of the above functional modules as an example when executing the posture detection method. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the posture detection device provided in the above embodiment and the posture detection method embodiment are based on the same concept. The implementation process is detailed in the method embodiment and will not be repeated here.

[0139] The serial numbers of the above embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0140] The present application also provides a computer storage medium that can store multiple instructions, which are suitable for being loaded and executed by a processor as described above. Figures 1-4 The posture detection method of the embodiment shown in the figure can be specifically executed by referring to Figures 1-4 The detailed description of the illustrated embodiment will not be repeated here.

[0141] The present application also provides a computer program product, which stores at least one instruction, and the at least one instruction is loaded and executed by the processor as described above. Figures 1-4 The posture detection method of the embodiment shown in the figure can be specifically executed by referring to Figures 1-4 The detailed description of the illustrated embodiment will not be repeated here.

[0142] Please refer to Figure 7 , which shows a block diagram of the structure of an electronic device provided by an exemplary embodiment of the present application. The electronic device in the present application may include one or more of the following components: a processor 110, a memory 120, an input device 130, an output device 140, and a bus 150. The processor 110, the memory 120, the input device 130, and the output device 140 may be connected via the bus 150.

[0143] The processor 110 may include one or more processing cores. The processor 110 utilizes various interfaces and circuits to connect various components within the electronic device. It executes instructions, programs, code sets, or instruction sets stored in the memory 120, as well as accesses data stored in the memory 120, to perform various functions and process data for the terminal 100. Optionally, the processor 110 may be implemented using at least one of the following hardware forms: a digital signal processing (DSP), a field-programmable gate array (FPGA), or a programmable logic array (PLA). The processor 110 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interfaces, and applications; the GPU is responsible for rendering and drawing display content; and the modem handles wireless communications. It is understood that the modem may not be integrated into the processor 110 and may be implemented separately via a communication chip.

[0144] The memory 120 may include a random access memory (RAM) or a read-only memory (ROM). Optionally, the memory 120 includes a non-transitory computer-readable storage medium. The memory 120 may be used to store instructions, programs, codes, code sets or instruction sets. The memory 120 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc. The operating system may be an Android system, including a system deeply developed based on the Android system, an iOS system developed by Apple, including a system deeply developed based on the iOS system or other systems.

[0145] The memory 120 can be divided into an operating system space and a user space. The operating system runs in the operating system space, and native and third-party applications run in the user space. In order to ensure that different third-party applications can achieve better operating results, the operating system allocates corresponding system resources to different third-party applications. However, the requirements for system resources in different application scenarios in the same third-party application are also different. For example, in the local resource loading scenario, the third-party application has higher requirements for disk reading speed; in the animation rendering scenario, the third-party application has higher requirements for GPU performance. The operating system and the third-party application are independent of each other, and the operating system often cannot perceive the current application scenario of the third-party application in a timely manner, resulting in the operating system being unable to perform targeted system resource adaptation according to the specific application scenario of the third-party application.

[0146] In order for the operating system to distinguish the specific application scenarios of third-party applications, it is necessary to open up data communication between third-party applications and the operating system so that the operating system can obtain the current scenario information of third-party applications at any time, and then perform targeted system resource adaptation based on the current scenario.

[0147] The input device 130 is used to receive input commands or data and includes, but is not limited to, a keyboard, a mouse, a camera, a microphone, or a touch-sensitive device. The output device 140 is used to output commands or data and includes, but is not limited to, a display device and a speaker. In one example, the input device 130 and the output device 140 may be combined, and the input device 130 and the output device 140 may be a touch-sensitive display.

[0148] The touch display screen can be designed as a full screen, a curved screen or a special-shaped screen. The touch display screen can also be designed as a combination of a full screen and a curved screen, or a combination of a special-shaped screen and a curved screen, which is not limited in the present embodiment.

[0149] In addition, those skilled in the art will understand that the structures of the electronic devices shown in the above figures do not limit the electronic devices. The electronic devices may include more or fewer components than shown, or may combine certain components, or arrange the components differently. For example, the electronic devices may also include radio frequency circuits, input units, sensors, audio circuits, wireless fidelity (WiFi) modules, power supplies, Bluetooth modules, and other components, which will not be described in detail here.

[0150] exist Figure 7 In the electronic device shown, the processor 110 may be configured to call a surface imaging application stored in the memory 120 and specifically perform the following operations:

[0151] A camera is used to collect the posture image of the target object;

[0152] Performing feature extraction processing on the posture image to obtain posture features of the target object;

[0153] Obtaining a target posture category label that matches the posture feature;

[0154] A target posture prompt voice is determined and output based on the target posture category label.

[0155] In one embodiment, the posture feature comprises at least one of a head posture feature, a shoulder posture feature, a back posture feature, a leg posture feature, and an overall posture feature.

[0156] In one embodiment, before obtaining the target posture category label that matches the posture feature, the processor 110 further performs the following operations:

[0157] Acquiring pressure sensing data of the target object using a pressure sensor;

[0158] When obtaining the target posture category label that matches the posture feature, the following operations are specifically performed:

[0159] Obtain a target posture category label that matches the pressure sensing data and the posture feature.

[0160] In one embodiment, the pressure sensing data includes at least one of front end pressure data, rear end pressure data, left end pressure data and right end pressure data on the seat surface of the target object seat.

[0161] In one embodiment, when the processor 110 performs feature extraction processing on the posture image to obtain the posture features of the target object, the processor 110 specifically performs the following operations:

[0162] inputting the posture image and the pressure sensing data into a posture recognition model;

[0163] Acquiring a posture feature of the object based on the posture recognition model;

[0164] When acquiring the target posture category label that matches the pressure sensing data and the posture feature, the following operations are specifically performed:

[0165] Obtain the target posture category label output by the posture recognition model.

[0166] In one embodiment, before using a camera to capture a posture image of the target object, the processor 110 further performs the following operations:

[0167] Creating an initial posture recognition model, obtaining a sample posture image, a sample posture category label corresponding to the sample posture image, and sample pressure sensing data corresponding to the sample posture image;

[0168] Inputting the sample posture image and the sample pressure sensing data into the initial posture recognition model to obtain a training posture category label output by the initial posture recognition model;

[0169] Based on the sample posture category labels and the training posture category labels, parameter adjustment processing is performed on the initial posture recognition type until model training is completed to obtain a posture recognition model.

[0170] In one embodiment, when the processor 110 determines and outputs the target posture prompt voice based on the target posture category label, it specifically performs the following operations:

[0171] Determining a target sitting posture state corresponding to the target posture category label, and obtaining a sitting posture state correction suggestion corresponding to the target sitting posture state;

[0172] Output a target posture prompt voice based on the sitting posture correction suggestion.

[0173] In this embodiment, an initial posture recognition model is created, sample posture images, sample posture category labels corresponding to the sample posture images, and sample pressure sensor data corresponding to the sample posture images are obtained. The sample posture images and sample pressure sensor data are input into the initial posture recognition model, and a training posture category label output by the initial posture recognition model is obtained. Based on the sample posture category label and the training posture category label, the parameters of the initial posture recognition type are adjusted until model training is completed, thereby obtaining a posture recognition model. Model training using sample posture images and sample pressure sensor data improves the accuracy and reliability of the posture recognition model, making the posture recognition model adaptable to different application environments and enhancing the model's generalization capabilities. A camera is used to capture posture images of a target object, and feature extraction is performed on the posture images to obtain posture features of the target object. These features are then extracted for different body parts of the target object. This multi-dimensional feature extraction reduces potential misjudgments caused by a single feature, while also addressing the individual characteristics of different subjects and improving the accuracy of sitting posture detection results. Pressure sensors are used to acquire pressure sensor data of the target object. Different pressure data can be obtained from pressure sensors located at different locations, allowing pressure conditions in different body parts of the target object to be detected. This facilitates multi-dimensional assessment of the target object's sitting posture, further reducing misjudgments. Posture images and pressure sensor data are input into a posture recognition model. The posture features of the subject are acquired based on the posture recognition model, and a target posture category label is output by the posture recognition model. The model's involvement in feature extraction and data analysis improves the efficiency of multimodal data processing and the accuracy of posture category label recognition. Combining pressure sensor data with posture features to determine the target subject's sitting posture, multimodal data fusion enhances the comprehensiveness of posture-related features, further improving posture detection accuracy. A target posture prompt is determined and output based on the target posture category label. By acquiring images and analyzing the target subject's sitting posture and providing voice prompts, the accuracy of abnormal sitting posture recognition is improved, enabling the subject to adjust their sitting posture in a timely manner, thereby ensuring normal skeletal development and preventing health problems caused by abnormal sitting posture. The target sitting posture corresponding to the target posture category label is determined, and posture correction suggestions corresponding to the target sitting posture are obtained. Based on the posture correction suggestions, a target posture prompt is output based on the posture correction suggestions. By outputting a target posture prompt voice containing the posture correction suggestions, the subject is provided with specific correction suggestions, making them easier to understand and implement.

[0174] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware through a computer program. The program can be stored in a computer-readable storage medium, and when executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory, or a random access memory.

[0175] The above disclosure is only a preferred embodiment of the present application, and certainly cannot be used to limit the scope of rights of the present application. Therefore, equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

[0176] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, storage, and display, etc.), and signals involved in the embodiments of this specification are all authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, the posture images and pressure sensor data involved in this specification are all obtained with full authorization.

Claims

1. A posture detection method, characterized in that: The method comprises: A camera is used to collect the posture image of the target object; Performing feature extraction processing on the posture image to obtain posture features of the target object; Obtaining a target posture category label that matches the posture feature; A target posture prompt voice is determined and output based on the target posture category label.

2. The method according to claim 1, characterized in that The posture feature includes at least one of a head posture feature, a shoulder posture feature, a back posture feature, a leg posture feature and an overall posture feature.

3. The method according to claim 1, characterized in that Before obtaining the target posture category label that matches the posture feature, the method further includes: Acquiring pressure sensing data of the target object using a pressure sensor; The obtaining of a target posture category label that matches the posture feature includes: Obtain a target posture category label that matches the pressure sensing data and the posture feature.

4. The method according to claim 3, characterized in that The pressure sensing data includes at least one of front end pressure data, rear end pressure data, left end pressure data and right end pressure data on the seat surface of the target object seat.

5. The method according to claim 3, characterized in that The performing feature extraction processing on the posture image to obtain the posture features of the target object includes: inputting the posture image and the pressure sensing data into a posture recognition model; Acquiring a posture feature of the object based on the posture recognition model; The acquiring of a target posture category label that matches the pressure sensing data and the posture feature includes: Obtain the target posture category label output by the posture recognition model.

6. The method according to claim 5, characterized in that Before the camera is used to collect the posture image of the target object, the method further includes: Creating an initial posture recognition model, obtaining a sample posture image, a sample posture category label corresponding to the sample posture image, and sample pressure sensing data corresponding to the sample posture image; Inputting the sample posture image and the sample pressure sensing data into the initial posture recognition model to obtain a training posture category label output by the initial posture recognition model; Based on the sample posture category labels and the training posture category labels, parameter adjustment processing is performed on the initial posture recognition type until model training is completed to obtain a posture recognition model.

7. The method according to claim 1, characterized in that The determining and outputting the target posture prompt voice based on the target posture category label includes: Determining a target sitting posture state corresponding to the target posture category label, and obtaining a sitting posture state correction suggestion corresponding to the target sitting posture state; Output a target posture prompt voice based on the sitting posture correction suggestion.

8. A posture detection device, characterized in that: The device comprises: An image acquisition unit, configured to acquire a posture image of a target object using a camera; a posture feature extraction unit, configured to perform feature extraction processing on the posture image to obtain posture features of the target object; A posture category acquisition unit, configured to acquire a target posture category label that matches the posture feature; A voice prompt unit is used to determine and output a target posture prompt voice based on the target posture category label.

9. A computer storage medium, characterized in that The computer storage medium stores a plurality of instructions, which are suitable for being loaded by a processor and executing the method steps according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: A processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the method steps according to any one of claims 1 to 7.

Images