Child safety seat adjustment method, device and system based on sitting position detection

Abstract

The present application relates to the technical field of automobile child safety seats, solves the problem that the abnormal sitting posture of a child sitting on a safety seat cannot be accurately detected in the prior art, and the probability of injury of the child during driving is high, and provides a child safety seat adjustment method, device and system based on sitting posture detection. The method comprises the following steps: acquiring a real-time video stream of a child safety seat area, decomposing the real-time video stream into multiple real-time images; performing image processing on each real-time image to output an abnormal sitting posture category; acquiring a preset adjustment strategy set of the child safety seat for abnormal sitting postures, and searching for a target strategy; if the target strategy exists, adjusting the state of the child safety seat according to the target strategy to match the sitting posture of the child; and if the target strategy does not exist, prompting the user to adjust the sitting posture of the child on the safety seat. The present application accurately detects abnormal sitting postures in a timely manner, and ensures the safety and comfort of the child during driving.

Description

Technical Field

[0001] This invention relates to the field of automotive child safety seat technology, and in particular to a child safety seat adjustment method, device, and system based on posture detection. Background Technology

[0002] In the early stages of a child's growth, their bodies are not yet fully developed, and their bones and muscles are still fragile. Therefore, an unstable seating arrangement can pose a serious threat to a child's health while the car is in motion. Child safety seats, as protective equipment specifically designed for them, play a crucial role.

[0003] However, the safety hazards posed by abnormal sitting postures of children during driving cannot be ignored, as these postures can directly impact a child's traffic safety. For example, firstly, if a child sits in a car seat in an abnormal posture such as leaning forward or sideways, the seatbelt's effectiveness in securing the child may be reduced. In the event of emergency braking or a collision, the seatbelt may not be able to secure the child tightly, exposing them to greater risks. Secondly, if a child is leaning forward in the seat, their head and neck may not receive sufficient support. This instability can lead to serious head and neck injuries, even life-threatening ones. Furthermore, car seats are designed for safety in a normal sitting posture; if a child sits in an abnormal posture, it may affect the seat's safety performance, preventing it from fully protecting the child in an accident. Abnormal sitting postures can also cause emotional instability in children, leading to anxiety and discomfort. This can distract them from the driving process, affecting the safety of both the driver and the child. In summary, abnormal sitting postures of children in car seats during driving can seriously endanger their safety. Correct sitting posture helps ensure that children are properly protected in the event of a vehicle collision or sudden braking. Parents and caregivers should always pay attention to the child's sitting posture to ensure that they maintain the correct posture in the car seat, thereby minimizing the risk of injury in traffic accidents.

[0004] Chinese patent CN115649025A discloses a device control method and apparatus for a child's abnormal posture in a cockpit. The method includes: acquiring the child's current first sleeping position in a smart child safety seat; determining whether the first sleeping position is an abnormal sleeping position, which includes at least one of the following: a sleeping position with a head tilt angle exceeding a preset head tilt angle, a sleeping position with a head tilt angle exceeding a preset head tilt angle, and a sleeping position with the head suspended in the air; if the first sleeping position is an abnormal sleeping position, querying a database of preset abnormal sleeping position adjustment sets for smart child safety seats, and determining whether the smart child safety seat has an auxiliary adjustment function for the first sleeping position; and then adjusting the smart child safety seat accordingly. When the smart child safety seat has an auxiliary adjustment function for the first sleeping position, it controls the child's current first sleeping position to be adjusted to a standard sleeping position according to the retrieved auxiliary adjustment strategy for the first sleeping position. When the smart child safety seat does not have an auxiliary adjustment function for the first sleeping position, it determines first control information based on the first sleeping position and sends the first control information to an output device. The output device prompts the user to adjust the child's first sleeping position to the target sleeping position, which is the standard sleeping position. The output device includes at least one of a speaker, a central control display, a passenger-side display, a rear-seat display, or a ceiling-mounted display controlled by a smart cockpit domain controller. However, on the one hand, detecting abnormal postures using a single parameter (head angle) is inaccurate; on the other hand, the above patent only applies to children sleeping in a safety seat and cannot be applied to detecting abnormal postures when children are awake.

[0005] Therefore, how to accurately detect abnormal sitting postures of children in car seats and adjust the car seats accordingly to avoid injury to children is an urgent problem to be solved. Summary of the Invention

[0006] In view of this, the present invention provides a child safety seat adjustment method, device and system based on posture detection, to solve the problem that the existing technology cannot accurately detect abnormal sitting postures of children sitting in safety seats, resulting in a high probability of injury to children during the journey.

[0007] The technical solution adopted in this invention is:

[0008] In a first aspect, the present invention provides a method for adjusting a child safety seat based on sitting posture detection, the method comprising:

[0009] S1: Acquire a real-time video stream of the child safety seat area and decompose the real-time video stream into multiple real-time images;

[0010] S2: Perform image processing on each of the real-time images, analyze whether the current sitting posture of the child sitting in the car seat is an abnormal sitting posture, and when it is determined to be an abnormal sitting posture, classify the abnormal sitting posture and output the abnormal sitting posture category.

[0011] S3: Obtain a preset set of adjustment strategies for abnormal sitting postures of child safety seats, and search for the target strategy corresponding to the abnormal sitting posture category in the set of adjustment strategies according to the abnormal sitting posture category.

[0012] S4: If the target strategy exists, adjust the child safety seat state according to the target strategy to match the child's sitting posture;

[0013] S5: If the target strategy does not exist, an alarm message will be issued to remind the user to adjust the child's sitting posture in the car seat.

[0014] Preferably, S2 includes:

[0015] S21: Obtain the preset child's age and corresponding age range;

[0016] S22: Determine whether the child is using the car seat for the first time, and based on the determination result and the age range information, obtain the standard sitting posture parameters;

[0017] S23: Obtain the real-time sitting posture parameters of the child in the car seat from the real-time image, compare the standard sitting posture parameters with the real-time sitting posture parameters, and output the abnormal sitting posture category.

[0018] Preferably, S22 includes:

[0019] S221: Obtain a preset usage time interval threshold. If it is determined to be the first use, or not the first use but the usage interval is longer than the threshold, then perform standard sitting posture detection on the current child and output the standard sitting posture parameters.

[0020] S222: If it is determined that the current child is not using the car seat for the first time and the usage interval is not greater than the duration threshold, then obtain the standard sitting posture parameters of the last use of the car seat.

[0021] Preferably, S221 includes:

[0022] S2211: Input the real-time images into a pre-trained deep learning model to obtain the posture angle information and position information corresponding to the preset parts of the child, wherein the preset parts include at least: head, shoulder, hand and knee.

[0023] S2212: Analyze and process the attitude angle information and position information, and output the attitude angle variance and position variance;

[0024] S2213: Compare the attitude angle variance and position variance with preset verification thresholds respectively. When the comparison result meets the verification threshold requirements, output the standard attitude angle information and standard position information at this time as the initial sitting posture parameters.

[0025] S2214: Obtain real-time feedback information from the user regarding the initial sitting posture parameters. When the user reports that the initial sitting posture parameters meet the requirements, output the current initial sitting posture parameters as the standard sitting posture parameters.

[0026] Preferably, S23 includes:

[0027] S231: Obtain the real-time posture angle information and real-time position information corresponding to the preset parts of the child in the real-time sitting posture parameters;

[0028] S232: Compare the standard attitude angle information with the real-time attitude angle information, and compare the standard position information with the real-time position information, and output the offset parameters corresponding to each preset part;

[0029] S233: Based on the weight coefficients corresponding to each preset part, perform weighted summation on each offset parameter and output the total offset;

[0030] S234: Compare the total offset with a preset offset threshold, and output the abnormal sitting posture category based on the comparison result.

[0031] Preferably, S234 includes:

[0032] S2341: Obtain the total target offset corresponding to multiple consecutive frames of images;

[0033] S2342: Compare the total number of each target offset with the offset threshold, and output the first number of times the total number of target offsets is greater than the offset threshold and the second number of times the total number of target offsets is not greater than the offset threshold;

[0034] S2343: Output the offset frequency based on the first and second counts;

[0035] S2344: Compare the offset frequency with a preset frequency threshold and output the abnormal sitting posture category.

[0036] Preferably, S4 includes:

[0037] S41: Obtain the target strategy in the set of adjustment strategies, wherein the target strategy includes at least one or more of the following strategies: head support height adjustment, seat height adjustment, seat tilt angle adjustment, and shoulder strap position adjustment;

[0038] S42: Based on the target strategy, control the child safety seat to adjust the sitting posture and obtain the adjusted target sitting posture parameters;

[0039] S43: Compare the target sitting posture parameters with the standard sitting posture parameters. If the offset frequency corresponding to the target sitting posture parameters is less than the frequency threshold, the adjustment is complete.

[0040] S44: If the offset frequency corresponding to the target sitting posture parameter is not less than the frequency threshold, a safety reminder is issued to the user.

[0041] Secondly, the present invention provides a child safety seat adjustment device based on sitting posture detection, the device comprising:

[0042] The image acquisition module is used to acquire a real-time video stream of the child safety seat area and decompose the real-time video stream into multiple real-time images;

[0043] The sitting posture detection module is used to process the real-time images and analyze whether the current sitting posture of the child sitting in the car seat is an abnormal sitting posture. When it is determined to be an abnormal sitting posture, the abnormal sitting posture is classified and the abnormal sitting posture category is output.

[0044] The strategy lookup module is used to obtain a preset set of adjustment strategies for abnormal sitting postures of the child safety seat, and to find the target strategy corresponding to the abnormal sitting posture category in the set of adjustment strategies according to the abnormal sitting posture category.

[0045] The seating posture adjustment module is used to adjust the state of the child safety seat according to the target strategy if the target strategy exists, so as to match the child's seating posture.

[0046] The safety warning module is used to issue an alarm message if the target strategy is not present, reminding the user to adjust the child's sitting posture in the car seat.

[0047] Thirdly, the present invention also provides a child safety seat adjustment system, the system comprising: a child safety seat body; a camera; and a controller; the child safety seat body is detachably installed in a vehicle; the camera is used to collect image data and transmit the collected image data to the controller for analysis and processing; the controller receives image data from the camera and analyzes the child's sitting posture; the system is used to implement the child safety seat adjustment method based on sitting posture detection as described above.

[0048] Preferably, the system further includes an alarm that receives detection results from the controller, the alarm including one or more of the following devices: a vehicle audio system, a central control display screen, a passenger-side display screen, a rear-seat display screen, or a ceiling-mounted display screen.

[0049] In summary, the beneficial effects of the present invention are as follows:

[0050] The present invention provides a child safety seat adjustment method, device, and system based on posture detection. The method includes: acquiring a real-time video stream of the child safety seat area and decomposing the real-time video stream into multiple real-time images; performing image processing on each real-time image to analyze whether the current sitting posture of the child sitting in the safety seat is an abnormal sitting posture; when an abnormal sitting posture is determined, classifying the abnormal sitting posture and outputting the abnormal sitting posture category; acquiring a preset set of adjustment strategies for child safety seats for abnormal sitting postures; searching for a target strategy corresponding to the abnormal sitting posture category in the set of adjustment strategies; if the target strategy exists, adjusting the state of the child safety seat according to the target strategy to match the child's sitting posture; if the target strategy does not exist, issuing an alarm message to remind the user to adjust the child's sitting posture in the safety seat. This invention acquires real-time video streams of the child safety seat area, enabling real-time monitoring of the child's posture and ensuring parents or guardians can constantly monitor the child's safety while the vehicle is in motion. Image processing technology rapidly analyzes real-time images to identify abnormal postures, providing warnings before the abnormal posture becomes severe to prevent potential dangers. Once an abnormal posture is detected, the child safety seat automatically adjusts its position according to a preset adjustment strategy to match the child's posture. This automated adjustment provides rapid response and enhanced safety. Different adjustment strategies are applied based on the type of abnormal posture, ensuring personalized care and tailored adjustments to each child's individual needs and differences. An alarm function alerts parents or guardians to posture problems, enhancing their safety awareness and helping them develop a habit of long-term vigilance and care for their children. Therefore, this invention helps improve children's safety and comfort in vehicles while providing parents and guardians with greater convenience and safety. However, in practical applications, privacy must be protected, data security ensured, and relevant laws and regulations must be followed. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, and these are all within the protection scope of the present invention.

[0052] Figure 1 This is a schematic diagram illustrating the overall working process of the child safety seat adjustment method based on posture detection in Embodiment 1 of the present invention;

[0053] Figure 2 This is a schematic diagram of the process for detecting a child's sitting posture in Embodiment 1 of the present invention;

[0054] Figure 3 This is a schematic diagram of the process for obtaining standard sitting posture parameters in Embodiment 1 of the present invention;

[0055] Figure 4 This is a schematic diagram of the process for detecting standard sitting posture during the first use in Embodiment 1 of the present invention;

[0056] Figure 5 This is a schematic diagram of the process for obtaining sitting posture detection results in Embodiment 1 of the present invention;

[0057] Figure 6 This is a schematic diagram of the process for detecting sitting posture based on offset frequency in Embodiment 1 of the present invention;

[0058] Figure 7 This is a schematic diagram of the process of adjusting the child safety seat in Embodiment 1 of the present invention;

[0059] Figure 8 This is a structural block diagram of the child safety seat adjustment device based on posture detection in Embodiment 2 of the present invention;

[0060] Figure 9 This is a schematic diagram of the child safety seat adjustment system based on posture detection in Embodiment 3 of the present invention;

[0061] The numbers in the diagram are as follows:

[0062] 1-Child safety seat body; 2-Camera; 3-Controller; 4-Alarm. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. In the description of the present invention, it should be understood that the terms "center," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, the element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. Where there is no conflict, embodiments of the present invention and the various features thereof can be combined with each other, all of which are within the scope of protection of the present invention.

[0064] Example 1

[0065] Please see Figure 1 Embodiment 1 of the present invention discloses a method for adjusting a child safety seat based on sitting posture detection, the method comprising:

[0066] S1: Acquire a real-time video stream of the child safety seat area and decompose the real-time video stream into multiple real-time images;

[0067] Specifically, the video stream of a child riding in a car seat is captured in real time by the in-vehicle camera on the child safety seat. This video stream consists of a series of continuous image frames, with multiple image frames captured per second. By processing the video stream frame by frame, each frame is extracted and formed into a continuous image sequence. Each frame is then analyzed and processed using image processing techniques. For example, based on image processing, computer vision techniques are used for posture detection, including human posture estimation, target detection, and key point localization, to accurately detect the child's posture.

[0068] S2: Perform image processing on each of the real-time images, analyze whether the current sitting posture of the child sitting in the car seat is an abnormal sitting posture, and when it is determined to be an abnormal sitting posture, classify the abnormal sitting posture and output the abnormal sitting posture category.

[0069] Specifically, attention is focused on the child safety seat area in the real-time image, avoiding interference from other parts of the vehicle and the environment. Using object detection algorithms or key part localization methods, key parts of the child, such as the head, shoulders, and torso, are located in the image to obtain the child's posture information. Based on the located key parts, computer vision algorithms analyze the position and posture angles of these parts to estimate the child's sitting posture. According to preset sitting posture standards (such as the relative position and angle of the head, spine, and waist), it is determined whether the sitting posture meets the standards. Based on the posture detection results, if an abnormal sitting posture is identified, the specific type of abnormality is further analyzed by comparing the child's actual sitting posture with... The pre-defined anomaly classification model subdivides abnormal sitting postures into different categories, such as head tilt, shoulder imbalance, and abnormal knee posture. For example, this anomaly classification model is based on a large amount of labeled sitting posture data and is trained using machine learning methods, such as Support Vector Machine (SVM) and Convolutional Neural Network (CNN), to more accurately classify abnormal sitting postures. Features such as the relative position, angle, and distance between key points are extracted as input features to the classification model, helping it better understand the characteristics of the sitting posture. The calculation and analysis steps are performed continuously in the decomposed real-time image stream to promptly detect and classify abnormal sitting postures, thereby outputting the abnormal sitting posture category. In summary, real-time monitoring can detect abnormal sitting postures in children early, preventing further deterioration. Automatic classification of abnormal sitting postures helps to automatically correct posture, providing a fast and accurate response without additional driver intervention. Different types of abnormal sitting postures can be accurately classified, allowing for appropriate adjustments to be made based on the specific type, ensuring children's comfort and safety.

[0070] In one embodiment, please refer to Figure 2 S2 includes:

[0071] S21: Obtain the preset child's age and corresponding age range;

[0072] Specifically, the system obtains the preset child's age and corresponding age range; based on the characteristics of each age range, it provides posture monitoring and reminders. For example, the first is the "infant" age range, which includes infants aged 0 to 1 year, followed by the "toddler" age range, which is suitable for children aged 1 to 3 years, and finally the "walking toddler" age range, which is suitable for toddlers aged 3 to 5 years.

[0073] S22: Determine whether the child is using the car seat for the first time, and based on the determination result and the age range information, obtain the standard sitting posture parameters;

[0074] Specifically, the system determines whether a child in each of the real-time images is using the car seat for the first time. If it is the child's first time using the car seat or if there has been a long interval, the child is identified as a new user. When this is determined, the parent or caregiver is guided to adjust the seat's angle, height, and other parameters to accommodate the child's body size and posture. Simultaneously, standard sitting posture parameters are obtained based on the child's age range. For example, for infants (0 to 1 year old): a more reclining and supportive seat setting is used to ensure adequate head and neck support, especially during the first use of the car seat. In this case, the parent or caregiver is guided... The angle and height of the car seat are adjusted according to the infant's body size and posture to ensure the seat meets their needs. For toddlers (1 to 3 years old): the angle and height of the seat may vary to accommodate their more stable and independent sitting posture. If it's the toddler's first time using the seat or a long interval has passed, adjustments are guided to ensure their comfort and safety. For toddlers (3 to 5 years old): toddlers are usually able to sit upright and have better posture control. In this case, the seat parameters may be closer to the standard setting, but their body size still needs to be considered. If it's their first time using the seat or a long interval has passed, adjustments are also guided accordingly. By comprehensively considering age ranges in step S22, the needs of children of different ages can be better met. Furthermore, the standard sitting posture parameters can be automatically adjusted as the child grows, ensuring they always enjoy optimal safety and comfort in the car seat. This personalized sitting posture monitoring and reminders help improve the child's riding experience, reduce the parental workload, and increase the child's safety in the vehicle. Conversely, if it's determined that the child is not using the seat for the first time and the interval is short, the previously determined standard sitting posture parameters are directly retrieved. By using facial recognition, personalized standard sitting posture parameters are set for each child, adapting to their body size and posture. For children who use the car seat frequently, the previously used posture parameters can be retrieved directly, reducing the workload for parents. Through this solution, combined with facial recognition, the system automatically determines whether to reset the standard sitting posture parameters based on the child's usage, thereby better ensuring the child's comfort and safety in the car seat.

[0075] In one embodiment, please refer to Figure 3 S22 includes:

[0076] S221: Obtain a preset usage time interval threshold. If it is determined to be the first use, or not the first use but the usage interval is longer than the threshold, then perform standard sitting posture detection on the current child and output the standard sitting posture parameters.

[0077] Specifically, images of children sitting in seats are captured periodically. These images undergo preprocessing, including noise removal and adjustments to brightness and contrast, to improve the accuracy of face recognition. Face detection algorithms, such as Haar cascade classifiers or deep learning methods, are used to process the images and identify facial regions. Then, key point localization techniques are used to mark important facial features, such as eyes, nose, and mouth, extracting feature vectors for each face. Deep learning models, such as convolutional neural networks (CNNs) or face embedding models, are typically used. These feature vectors are used for subsequent comparison and recognition. Registered children's facial information, including embedded feature vectors and related identification identifiers, is stored in a database. This facial information can be compared with real-time captured face images. When new face images are captured, the database is updated accordingly. The system compares the previously embedded feature vectors with data in the database to identify the child in the current image. If no match is found in the database, it is confirmed as the child's first use; otherwise, it is considered a non-first use. If it is determined to be a first use, or a non-first use but the current usage interval is less than the specified duration threshold, standard sitting posture detection is triggered. The system uses a camera to acquire an image of the child and analyzes the relative positions and angles of key parts such as the child's head and torso using computer vision technology to determine standard sitting posture parameters. These parameters include at least the head tilt angle and the degree of torso bending. The standard sitting posture parameters are stored in the system for future comparison and adjustment. The standard sitting posture parameters can be used for real-time monitoring. The actual sitting posture of the child is compared with the standard sitting posture parameters. If an anomaly is detected, the user will be alerted to make adjustments.

[0078] In one embodiment, please refer to Figure 4 S221 includes:

[0079] S2211: Input the real-time images into a pre-trained deep learning model to obtain the posture angle information and position information corresponding to the preset parts of the child, wherein the preset parts include at least: head, shoulder, hand and knee.

[0080] Specifically, an image dataset related to the use of child car seats is collected. This data should include images of children in different postures, angles, and lighting conditions, ensuring that the dataset contains various seat shapes and postures to reflect actual usage scenarios. A suitable base model for the object detection task is selected, such as YOLO, Faster R-CNN, SSD, etc. These models are all pre-trained general models, and fine-tuning is performed on them. Pre-trained model weights are used as initial weights, which are trained on a large-scale dataset as the starting point for the model to reduce the training time required for fine-tuning. According to the specific needs of car seats, the object detection model is fine-tuned, which requires modifying the network structure, such as changing the number of classes in the classifier, adjusting the size of the input image, or adding some custom layers to adapt to the shape and posture of the seat. Data augmentation is performed to expand the training set, including image flipping, rotation, cropping, adjusting brightness and contrast, etc., to increase the robustness and generalization ability of the model. The loss function of the model is adjusted according to the application scenario of the child car seat and the preset target parts of the child. For example, the weights of classification loss and bounding box regression loss are adjusted to make the model focus more on specific preset body parts, including at least the head, shoulders, hands, and knees. Using a prepared dataset and a modified model, fine-tuning training is performed. During training, image data from car safety seats is used to update the model's weights, gradually adapting it to specific scenarios. The fine-tuned model is evaluated using a validation or test set to check its performance in child safety seat scenarios, with evaluation metrics including accuracy, recall, and precision. Based on the evaluation results, the model is adjusted and iterated as needed, requiring multiple fine-tuning and evaluations to achieve better performance. After fine-tuning, the model is applied to actual car safety seats, and its performance is verified through testing in child safety seats. A camera is installed above the headrest of the entire seat, allowing it to look down at the child's head and upper body and capture clear images of key areas such as the head, shoulders, and hands. This placement avoids obstruction from other parts of the seat. Cameras are also installed on the sides of the seat, close to the child, to capture their posture, particularly the hands and knees. The camera's angle and view need to match the shape of the seat's side to obtain clear side images. Placing multiple cameras in different locations and capturing images from different angles, then fusing the image data, can provide more comprehensive information about the child's body parts. The choice of camera placement should be based on the seat's design, the child's posture, and how the seat will be used.The camera placement should ensure clear capture of key areas, avoiding obstructions and shadows, while providing sufficient image quality to support subsequent part localization and pose recognition tasks. The model is used to analyze real-time images captured by cameras at different locations to obtain the posture angle information and position information corresponding to preset parts of the child in the real-time images. The posture angle information and position information include numerical values ​​of posture angles, such as the tilt angle of the head and the rotation angle of the shoulders, as well as the preset coordinate positions of the child's key parts on the image.

[0081] S2212: Analyze and process the attitude angle information and position information, and output the attitude angle variance and position variance;

[0082] Specifically, for preset key body parts (such as head, shoulders, hands, and knees), the corresponding posture angle information is extracted from the model output. The mean (μ) of these angles and the square of the difference between each angle and the mean ((θ-μ)^2) are calculated. Then, the average of the squares of these differences is calculated, which is the posture angle variance. Similarly, the position information (coordinates) of the preset body parts is extracted from the model output. The mean (μ) of these coordinates and the square of the difference between each coordinate and the mean ((x-μ)^2) are calculated. Then, the average of the squares of these differences is calculated, which is the position variance. The values ​​of posture angle variance and position variance are obtained. These values ​​represent the degree of dispersion of posture angles and positions. The smaller the variance, the more stable and concentrated the posture angles and positions are. By comparing the magnitudes of posture angle variance and position variance, the stability of different key body parts can be analyzed. Smaller variance may indicate that the child's sitting posture is relatively stable, while larger variance may mean that the sitting posture is unstable or that the position of the child's key body parts fluctuates greatly.

[0083] S2213: Compare the posture angle variance and position variance with preset verification thresholds respectively. When the comparison results meet the verification threshold requirements, output the standard posture angle information and standard position information as initial sitting posture parameters. Specifically, compare the calculated posture angle variance with the preset verification threshold. If the posture angle variance is less than or equal to the threshold, it indicates that the child's posture angle is stable and meets the requirements. Similarly, compare the calculated position variance with the preset verification threshold. If the position variance is less than or equal to the threshold, it indicates that the child's key body parts are stable and meet the requirements. If both the posture angle variance and position variance are less than or equal to the corresponding threshold, that is, both posture stability and position stability meet the preset requirements, then the child's sitting posture can be considered stable at this time, and the posture angle information and position information at this time are used as standard sitting posture parameters. Based on the judgment results, output the standard posture angle information and standard position information that meet the requirements as initial sitting posture parameters. These standard parameters can be used for subsequent comparison and adjustment to ensure that the child maintains the correct sitting posture in the car seat.

[0084] S2214: Obtain real-time feedback information from the user regarding the initial sitting posture parameters. When the user reports that the initial sitting posture parameters meet the requirements, output the current initial sitting posture parameters as the standard sitting posture parameters.

[0085] Specifically, the initial sitting posture parameters are displayed to the user, for example, through the smart display interface on the car seat or a mobile application. Real-time feedback from the user after seeing these parameters is obtained, including clicking the confirm or reject button on the touchscreen. Based on the user's feedback, it is determined whether the initial sitting posture parameters meet the requirements. If the user confirms that the sitting posture parameters are appropriate, these initial parameters are output as the standard sitting posture parameters. Using the standard sitting posture parameters: User participation in setting the sitting posture parameters ensures that they are closer to actual needs, improving the child's sitting comfort and safety. User feedback avoids overly theoretical parameter settings, making them more consistent with real-world usage.

[0086] S222: If it is determined that the current child is not using the car seat for the first time and the usage interval is not greater than the duration threshold, then obtain the standard sitting posture parameters of the last use of the car seat.

[0087] Specifically, if the facial recognition system detects that the child is not using the car seat for the first time, it retrieves the time interval between the child's last use of the car seat. For example, if the child last used the car seat 3 days ago, the time interval is 3 days. Assuming the preset time threshold is 7 days, and the time interval is less than 7 days, it retrieves the standard sitting posture parameters from the last use of the car seat. This allows the previously determined sitting posture parameters to be used directly when the car seat is used again in a short period. For cases of frequent car seat use, the previously determined standard sitting posture parameters are used directly, eliminating the need for repeated standard sitting posture detection, thus saving time and providing greater convenience.

[0088] S23: Obtain the real-time sitting posture parameters of the child in the car seat from the real-time image, compare the standard sitting posture parameters with the real-time sitting posture parameters, and output the abnormal sitting posture category.

[0089] In one embodiment, S23 includes:

[0090] S231: Preprocess the real-time image, use a pre-trained deep learning model to extract features from the preprocessed image, and output feature information;

[0091] S232: Using a key point localization method or a human posture estimation model, analyze the feature information and output the posture angle and position of the child's preset parts as the real-time sitting posture parameters.

[0092] S233: Compare the standard parameters of each preset part in the standard sitting posture parameters with the real-time sitting posture parameters one by one, and output the abnormal sitting posture category.

[0093] Specifically, after determining the standard sitting posture parameters, the images captured in real-time by the camera during subsequent driving are preprocessed, including scaling, normalization, and noise reduction, to ensure image quality and consistency and prepare for subsequent feature extraction. A pre-trained deep learning model, such as a convolutional neural network (CNN), is used to extract features from the preprocessed images. The last layer of the model obtains feature representations of various parts of the image. Based on the extracted features, keypoint localization methods or human pose estimation models, such as OpenPose, can be used to estimate the positions of the child's head, shoulders, hands, and knees. Using the estimated positions, the posture is calculated... The system integrates and outputs real-time sitting posture parameters, including the angles and positions of the head, shoulders, hands, and knees, relative to a preset sitting posture, such as the head tilt angle and shoulder rotation angle. These real-time posture parameters reflect the child's real-time sitting posture. Standard sitting posture parameters (standard angles and positions of the head, shoulders, hands, and knees) are compared with the real-time parameters one by one. For each posture angle and position, a pre-set threshold range is used to determine whether it meets the standard. For example, a maximum allowable tilt angle range is set for the head tilt angle. For each part, based on the comparison between the real-time and standard sitting posture parameters, it is determined whether the current sitting posture is within the allowable range. If the posture angles and positions of all parts are within the allowable threshold range, it is determined to meet the standard sitting posture; otherwise, it is determined to be an abnormal sitting posture. Based on the determination result, the sitting posture detection result is output. If the sitting posture meets the standard, the result "Standard Sitting Posture" is output; if any part does not meet the standard, the result "Abnormal Sitting Posture" is output, and a detailed description of the abnormality may be provided. Child safety seats are designed with the car's internal structure in mind, providing appropriate support and protection for children. They compare standard and real-time seating parameters to ensure a comfortable and correct posture for children while maintaining safety. Correct posture is a key factor in ensuring children's safety during driving. By comparing real-time and standard postures, abnormal postures, such as head tilting or torso bending, can be detected, promptly alerting parents or guardians for adjustments to ensure the child's safety and stability. Incorrect posture can lead to discomfort or even adverse effects on children during driving, such as neck fatigue and general discomfort. Real-time monitoring and comparison of posture can prevent potential adverse effects, ensuring a more comfortable and safer riding experience for children. Embedding posture parameter comparison into the child safety seat system achieves an organic integration of hardware and software. The hardware captures real-time images through a rational structural design and camera layout; the software uses deep learning models to extract and analyze features from the images, outputting posture detection results. Working together, they provide comprehensive protection for children while traveling.

[0094] In one embodiment, please refer to Figure 5 S23 includes:

[0095] S231: Obtain the real-time posture angle information and real-time position information corresponding to the preset parts of the child in the real-time sitting posture parameters;

[0096] S232: Compare each of the standard attitude angle information with each of the real-time attitude angle information, and compare each of the standard position information with each of the real-time position information, and output the offset parameters corresponding to each preset part;

[0097] Specifically, firstly, for preset body parts, such as the head, shoulders, hands, and knees, corresponding posture angles and position information are extracted from real-time images using technologies such as facial recognition and target detection. This information reflects the actual situation of the child currently sitting in the car seat. The extracted real-time posture angle information is compared with preset standard posture angle information, and offset parameters are output. These offset parameters include the tilt angle of the head, the angle of the shoulders, the position of the hands, and the degree of knee flexion. By calculating the difference between the actual posture and the standard posture, posture offset parameters for each preset body part are obtained. At the same time, the real-time position information is compared with preset standard position information to determine the child's specific position in the seat, whether they are sitting upright, and the position of their waist and knees. By comparing the actual position with the standard position, and considering factors such as whether the child is properly supported, the system obtains positional offset parameters for each preset part. By comparing real-time posture angle and position information with preset standards and calculating the posture and position offset parameters, the system accurately assesses the difference between the child's sitting posture and the expected standard. This effectively reflects the child's sitting posture in actual car seat use and provides parents or guardians with information on whether the child's posture needs adjustment. This feedback helps ensure the correct sitting posture of children in car seats, thereby improving safety and comfort and reducing potential risks. In summary, comparing posture angle and position information helps improve the actual use effect of car seats and ensures the safety and comfort of children.

[0098] S233: Obtain the weight coefficients corresponding to each preset part, perform weighted summation on each offset parameter, and output the total offset;

[0099] Specifically, considering the offset parameters of various parts of the child's body, the importance of different parts and their impact on sitting stability and safety are taken into account. The offset of different parts will have different effects on the child's comfort and safety. Therefore, it is necessary to perform weighted summation on the offset parameters of different parts to obtain the total offset. In the weighted summation, a weight can be assigned to each preset part to reflect its importance in sitting stability and safety. For example, the offset of the head may put a greater burden on the neck, leading to an increased risk of neck discomfort or injury. Therefore, the weight of the head part can be higher, while the offset of the hands may have a relatively smaller impact on the overall sitting stability. Therefore, the weight of the hands part can be lower. For example, consider a situation where a large head offset angle exceeds the safe range, which could lead to excessive neck twisting and increase the risk of neck injury. On the other hand, a slight shoulder offset might affect the balance of the sitting posture but would not pose a direct threat to the child's safety. Therefore, when weighting the sum, the head is given a higher weight. When weighting the offset parameters of each preset part, the total offset can be calculated using the following formula: Total offset = Σ(weight × offset parameter), where Σ represents the summation over all preset parts, the weight is the weight value corresponding to each part, and the offset parameter is the actual offset value of each part. Suppose we have four preset parts: head, shoulder, hand, and knee. We assign different weights to different preset body parts to reflect their importance: head weight: 0.4; shoulder weight: 0.3; hand weight: 0.2; knee weight: 0.1. Assuming the actual offset angles and positions of each body part in the real-time image are: head offset angle: 5°; shoulder offset angle: 2°; hand offset angle: 1°; knee offset angle: 0.5°, based on the above weights and actual offset parameters, the total offset is calculated as follows: Total offset = (0.4 × 5°) + (0.3 × 2°) + (0.2 × 1°) + (0.1 × 0.5°) = 2.7°. This total offset of 2.7° reflects the overall degree of offset between the child's sitting posture and the standard sitting posture. By obtaining the overall total offset, we reflect the difference between the child's sitting posture and the standard sitting posture, considering the importance and influence of each body part. This total offset is used as a further decision indicator to determine whether the current sitting posture meets safety requirements.

[0100] S234: Compare the total offset with a preset offset threshold, and output the abnormal sitting posture category based on the comparison result.

[0101] Specifically, if the total offset exceeds the preset offset threshold, it means that the current sitting posture may pose a safety hazard, and the child's sitting posture needs to be adjusted in time to ensure their comfort and safety in the car seat. Overall, the weighted summation process can more accurately measure the impact of the offset of each part of the child on the sitting posture, helping users to better understand and judge the child's sitting posture and take appropriate adjustment measures to ensure the child's safety and comfort in the car seat.

[0102] In one embodiment, please refer to Figure 6 S234 includes:

[0103] S2341: Obtain the total target offset corresponding to multiple consecutive frames of images;

[0104] Specifically, the total target offset corresponding to multiple consecutive frames of images is obtained; the total offset in multiple consecutive frames of images is taken into account to obtain a more accurate sitting posture offset. In the application of car child safety seats, multiple frames of images can capture the changes in a child's sitting posture over a period of time, thereby more comprehensively assessing the stability of the sitting posture. By calculating the total offset in each frame of images and summing them up, the total target offset corresponding to multiple consecutive frames of images is obtained.

[0105] S2342: Compare the total number of each target offset with the offset threshold, and output the first number of times the total number of target offsets is greater than the offset threshold and the second number of times the total number of target offsets is not greater than the offset threshold;

[0106] Specifically, the total target offset is compared with the offset threshold, and the first number of times the total target offset exceeds the offset threshold and the second number of times the total target offset does not exceed the offset threshold are output. This step aims to analyze the distribution of the total target offset and further evaluate the stability of the sitting posture. By comparing the total target offset with the set offset threshold, it is possible to determine which frames have offsets exceeding the threshold and which frames have offsets not exceeding the threshold. The first count represents the number of frames in a series where the offset exceeds the threshold for the first time, and the second count represents the number of frames in a series where the offset does not exceed the threshold. This analysis helps to identify temporary offset changes and sustained stable sitting postures.

[0107] S2343: Output the offset frequency based on the first and second counts;

[0108] Specifically, based on the first and second counts, the offset frequency is output. By calculating the offset frequency, that is, the proportion of offsets occurring in multiple consecutive frames, the offset frequency can be used to determine the stability of the sitting posture. The offset frequency can be obtained by dividing the first count by the total number of frames, which represents the proportion of offsets exceeding a threshold in multiple frames of images. The level of the offset frequency can help determine the continuity and changes in the sitting posture.

[0109] S2344: Compare the offset frequency with a preset frequency threshold and output the abnormal sitting posture category.

[0110] Specifically, the offset frequency is compared with a preset frequency threshold to output the abnormal sitting posture category. In this step, the calculated offset frequency is compared with the preset frequency threshold. If the offset frequency exceeds the preset threshold, it means that the child's sitting posture has changed significantly and may be unstable. Conversely, if the offset frequency is within an acceptable range, the child's sitting posture may be relatively stable. Based on the comparison result, a corresponding sitting posture detection result can be output to remind the user whether they need to adjust their sitting posture to maintain a stable and safe state.

[0111] S3: Obtain a preset set of adjustment strategies for abnormal sitting postures of child safety seats, and search for the target strategy corresponding to the abnormal sitting posture category in the set of adjustment strategies according to the abnormal sitting posture category.

[0112] Specifically, a preset set of adjustment strategies for abnormal sitting postures in child safety seats is obtained. This preset set of adjustment strategies includes various methods for adjusting abnormal sitting postures, covering aspects such as the head, shoulders, hands, and knees, to address different types of sitting posture problems. Based on previous abnormal sitting posture detection and classification results, the abnormal sitting posture category is associated with the corresponding adjustment method in the preset set of adjustment strategies. For example, head tilt requires adjustment of head support, and shoulder imbalance requires adjustment of shoulder support. Once an abnormal sitting posture is detected and classified, a target strategy corresponding to the abnormal sitting posture category is searched in the preset set of abnormal sitting posture adjustment strategies. This target strategy is the method actually used to adjust the child safety seat. By combining abnormal sitting posture classification with the set of adjustment strategies, the child's sitting posture problems can be corrected more accurately and effectively, improving the child's safety and comfort in the vehicle.

[0113] S4: If the target strategy exists, adjust the child safety seat state according to the target strategy to match the child's sitting posture;

[0114] Specifically, if a target strategy is successfully found during the previous strategy search process in an abnormal sitting posture, the child safety seat is controlled according to this target strategy to adjust the sitting posture. Based on the selected target strategy, corresponding adjustment instructions are generated. These instructions involve the position, tilt angle, and support tightness of the child safety seat components, etc. The adjustment instructions are sent via a communication protocol, and the safety seat will automatically adjust itself according to the received instructions. For example, adjusting the seat angle and the height of the headrest.

[0115] In one embodiment, please refer to Figure 7 S4 includes:

[0116] S41: Obtain the target strategy in the set of adjustment strategies, wherein the target strategy includes at least one or more of the following strategies: head support height adjustment, seat height adjustment, seat tilt angle adjustment, and shoulder strap position adjustment;

[0117] Specifically, a target strategy is obtained from a preset set of adjustment strategies. This target strategy is set for different situations of abnormal sitting posture and includes at least the following: head support height adjustment: by adjusting the height of the head support, the head is properly supported, preventing the head from tilting or being too high; seat height adjustment: by adjusting the height of the seat, the child's feet can be firmly placed on the seat, keeping the knees at the correct angle; seat tilt angle adjustment: by adjusting the tilt angle of the seat, the child's torso is properly supported, preventing the torso from leaning forward or backward; shoulder strap position adjustment: by adjusting the position of the seat belt, the seat belt is properly secured to the child's shoulders and waist, preventing them from slipping or twisting during travel.

[0118] S42: Based on the target strategy, control the child safety seat to adjust the sitting posture and obtain the adjusted target sitting posture parameters;

[0119] Specifically, after selecting a target strategy based on the classification of abnormal sitting postures and the set of adjustment strategies, the requirements for target sitting posture parameters are defined according to the target strategy. These parameters may include the height of the head support, the tilt angle of the seat, the position of the shoulder straps, etc. Based on the set target sitting posture parameters, different components of the child safety seat, such as the head support, the tilt mechanism of the seat, and the fixing position of the shoulder straps, are controlled to adjust the seat to a suitable position and angle. For the position-related parts of the target sitting posture parameters, the height and tilt angle of the seat are controlled through electric or mechanical mechanisms to ensure that the child's body parts are properly supported. For the posture angle parts of the target sitting posture parameters, the mechanical structure of the seat is controlled to ensure that the child's body is in the correct posture, such as the angle between the head and torso and the degree of knee flexion. During the adjustment process, the child's actual sitting posture is monitored in real time to adjust the seat in a timely manner and ensure the accuracy of the adjustment. After the adjustment is completed, the adjusted target sitting posture parameters are generated to verify whether the adjustment effect has achieved the expected results.

[0120] S43: Compare the target sitting posture parameters with the standard sitting posture parameters. If the offset frequency corresponding to the target sitting posture parameters is less than the frequency threshold, the adjustment is complete.

[0121] Specifically, the adjusted target sitting posture parameters are compared with the standard sitting posture parameters item by item. By calculating the offset value of each sitting posture parameter, the offset frequency is obtained. The offset frequency refers to the frequency of the offset between the real-time sitting posture parameter and the standard parameter within a certain period of time. The calculated offset frequency is compared with a preset frequency threshold, which represents the upper limit of the offset frequency. If the offset frequency of the target sitting posture parameter is less than the set frequency threshold, the adjustment is considered complete and corresponding marking is performed. Considering that each child may have different sitting posture needs, the system can support personalized frequency threshold settings to meet the needs of different users.

[0122] S44: If the offset frequency corresponding to the target sitting posture parameter is not less than the frequency threshold, a safety reminder is issued to the user.

[0123] Specifically, if the comparison process reveals that the frequency of deviation from the target sitting posture parameters after system adjustment exceeds a preset frequency threshold, indicating a significant difference between the child's sitting posture and the standard sitting posture, a safety reminder will be issued to the user. Safety reminders can be delivered through prominent methods such as alarm sounds or flashing lights to attract the user's attention. Alternatively, text messages can be displayed on the screen informing the user that the child's posture is problematic and requires further adjustment. Vibration can also be used to emphasize the necessity of posture adjustment. In the event of a safety reminder, the user may need to make additional manual adjustments to bring the posture closer to the standard. The interval between reminders can be set as needed to avoid frequent user interruptions. The safety reminder notification can also include suggestions on how to adjust the chair to achieve a more suitable posture. Each safety reminder will be recorded for subsequent analysis and improvement.

[0124] S5: If the target strategy does not exist, an alarm message will be issued to remind the user to adjust the child's sitting posture in the car seat.

[0125] Specifically, if the target strategy is not present, a safety reminder is issued to the user, prompting them to adjust the child's seating position. Within the car, this safety reminder can be implemented in several ways to ensure the child's safety, such as: audible alerts: An audible alert is issued through the car audio system or the child safety seat's built-in speaker to remind the driver or passengers of the child's abnormal seating position; this could be an audible warning, voice prompt, or alarm tone. Visual alerts: Warning messages, such as "Please adjust the child's seating position," are displayed on the car's central control screen, passenger-side screen, rear-seat screen, or ceiling-mounted screen; this directly attracts the driver's or passenger's attention. Vibration alerts: Vibration signals are transmitted to the driver or passengers using the car seat's vibration function, drawing their attention without distracting them. Mobile phone notifications: If the in-vehicle system is connected to a mobile phone, a notification is sent to the phone to remind the driver or passengers to adjust the child's seating position; this is very useful when the driver cannot immediately notice the warning. Voice assistant prompts: If the car is equipped with a voice assistant (such as Siri, Google Assistant, etc.),... Multiple reminder methods can be implemented, including: voice alerts (such as assistants); light alerts (LED lights in certain areas of the car that change color or flashing pattern to alert the driver or passengers when posture is detected as abnormal); seat vibration alerts (vibrating seats to alert the driver or passengers to abnormal posture); and mobile app alerts (notifications sent to the driver or passengers' phones to check the child's posture). These methods, combined with the car's internal structure, ensure timely attention to the child's posture even when the driver is focused on driving, thus protecting the child's safety. Choosing the appropriate alert method requires considering the driver's or passenger's perception and preferences, and effectively conveying warning information without interfering with driving.

[0126] Example 2

[0127] Please see Figure 8 Embodiment 2 of the present invention also provides a child safety seat adjustment device based on sitting posture detection, the device comprising:

[0128] The image acquisition module is used to acquire a real-time video stream of the child safety seat area and decompose the real-time video stream into multiple real-time images;

[0129] The sitting posture detection module is used to process the real-time images and analyze whether the current sitting posture of the child sitting in the car seat is an abnormal sitting posture. When it is determined to be an abnormal sitting posture, the abnormal sitting posture is classified and the abnormal sitting posture category is output.

[0130] In one embodiment, the sitting posture detection module includes:

[0131] The child age analysis unit is used to obtain the preset child age and corresponding age range;

[0132] The standard sitting posture parameter acquisition unit is used to determine whether the child is using the car seat for the first time, and to acquire the standard sitting posture parameters based on the determination result and the age range information.

[0133] In one embodiment, the standard sitting posture parameter acquisition unit includes:

[0134] The standard sitting posture detection subunit is used to obtain a preset usage time interval threshold. If it is determined to be the first use, or not the first use but the usage interval is longer than the threshold, then the standard sitting posture is detected for the current child and the standard sitting posture parameters are output.

[0135] In one embodiment, the standard sitting posture detection subunit includes:

[0136] The preset body part detection subunit is used to input the real-time images into a pre-trained deep learning model to obtain the posture angle information and position information corresponding to the preset body parts of the child. The preset body parts include at least: head, shoulder, hand and knee.

[0137] The attitude angle and position analysis subunit is used to analyze and process the attitude angle information and position information, and output the attitude angle variance and position variance.

[0138] The verification subunit is used to compare the attitude angle variance and position variance with the corresponding verification thresholds respectively. When the comparison result meets the verification threshold requirements, the standard attitude angle information and standard position information at this time are output as the initial sitting posture parameters.

[0139] The secondary verification subunit is used to obtain real-time feedback information from the user on the initial sitting posture parameters. When the user reports that the initial sitting posture parameters meet the requirements, the initial sitting posture parameters at this time are output as the standard sitting posture parameters.

[0140] The standard sitting posture parameter acquisition subunit is used to acquire the standard sitting posture parameters of the last time the child used the car seat if it is determined that the current child is not using the car seat for the first time and the usage interval is not greater than the duration threshold.

[0141] The posture comparison unit is used to acquire the real-time posture parameters of a child in a car seat in a real-time image, compare the standard posture parameters with the real-time posture parameters, and output the abnormal posture category.

[0142] In one embodiment, the sitting posture comparison unit includes:

[0143] The real-time sitting posture information acquisition subunit is used to acquire the real-time posture angle information and real-time position information corresponding to the preset parts of the child in the real-time sitting posture parameters.

[0144] The offset parameter calculation subunit is used to compare each of the standard attitude angle information with each of the real-time attitude angle information, and to compare each of the standard position information with each of the real-time position information, and output the offset parameter corresponding to each preset part.

[0145] The weighted processing subunit is used to obtain the weight coefficients corresponding to each preset part, perform weighted summation on each of the offset parameters, and output the total offset.

[0146] The detection result output subunit is used to compare the total offset with a preset offset threshold, and output the abnormal sitting posture category based on the comparison result.

[0147] In one embodiment, the detection result output subunit includes:

[0148] The target offset total calculation subunit is used to obtain the target offset total corresponding to multiple consecutive frames of images;

[0149] The offset comparison subunit is used to compare the total offset of each target with the offset threshold, and output the first number of times the total offset of the target is greater than the offset threshold and the second number of times the total offset of the target is not greater than the offset threshold;

[0150] The offset frequency calculation subunit is used to output the offset frequency based on the first count and the second count;

[0151] The posture detection result output subunit is used to compare the offset frequency with a preset frequency threshold and output the abnormal posture category.

[0152] The strategy lookup module is used to obtain a set of preset adjustment strategies for abnormal sitting postures of child safety seats, and to find the target strategy corresponding to the abnormal sitting posture category in the set of adjustment strategies based on the abnormal sitting posture category.

[0153] The seating posture adjustment module is used to adjust the state of the child safety seat according to the target strategy if the target strategy exists, so as to match the child's seating posture.

[0154] In one embodiment, the posture adjustment module includes:

[0155] The target strategy acquisition unit is used to acquire the target strategy in the set of adjustment strategies, wherein the target strategy includes at least one or more of the following strategies: head support height adjustment, seat height adjustment, seat tilt angle adjustment, and shoulder strap position adjustment.

[0156] The real-time sitting posture adjustment unit is used to control the child safety seat to adjust the sitting posture according to the target strategy and obtain the adjusted target sitting posture parameters.

[0157] The adjustment completion confirmation unit is used to compare the target sitting posture parameters with the standard sitting posture parameters. If the offset frequency corresponding to the target sitting posture parameters is less than the frequency threshold, the adjustment is complete.

[0158] The safety warning unit is used to issue a safety reminder to the user if the offset frequency corresponding to the target sitting posture parameter is not less than the frequency threshold.

[0159] The safety reminder module is used to issue an alarm message if the target policy is not present, reminding the user to adjust the child's sitting posture in the car seat.

[0160] Specifically, the child safety seat adjustment device based on posture detection provided in this embodiment of the invention includes: an image acquisition module for acquiring a real-time video stream of the child safety seat area and decomposing the real-time video stream into multiple real-time images; a posture detection module for performing image processing on each of the real-time images, analyzing whether the current posture of the child sitting in the safety seat is an abnormal posture, and when an abnormal posture is determined, classifying the abnormal posture and outputting the abnormal posture category; a strategy search module for acquiring a preset set of adjustment strategies for the child safety seat for abnormal postures, and searching for a target strategy corresponding to the abnormal posture category in the set of adjustment strategies; a posture adjustment module for adjusting the state of the child safety seat to match the child's posture if the target strategy exists; and a safety warning module for issuing an alarm message to remind the user to adjust the child's posture in the safety seat if the target strategy does not exist. This device monitors a child's posture in a car seat in real time and provides rapid feedback. If an abnormal posture is detected, the user can take timely measures to ensure the child's safety and comfort. Through image processing technology and a set of adjustment strategies, it automatically determines whether the posture is abnormal and adjusts it according to preset strategies. This automated adjustment avoids manual intervention and provides a more accurate and timely response. The set of adjustment strategies provides personalized adaptation for different types of abnormal postures; different posture problems may require different adjustment methods, allowing the selection of an appropriate adjustment strategy based on the detected abnormality type. During long drives, drivers and passengers may overlook a child's posture due to fatigue or inattention. This invention reduces such human negligence, ensuring the child's safety and comfort while driving.

[0161] Example 3

[0162] Please see Figure 9Embodiment 3 of the present invention also provides a child safety seat adjustment system, the system comprising: a child safety seat body 1; a camera 2; and a controller 3; the child safety seat body 1 is detachably installed in the vehicle; the camera 2 is used to collect image data and transmit the data to the controller 3 for analysis and processing; the controller 3 receives image data from the camera 2 and analyzes the child's sitting posture; the system is used to implement the child safety seat adjustment method based on sitting posture detection as described in Embodiment 1.

[0163] In one embodiment, the system further includes an alarm 4 that receives detection results from a controller, and the alarm includes one or more of the following devices: a vehicle audio system, a central control display screen, a passenger-side display screen, a rear-seat display screen, or a ceiling-mounted display screen.

[0164] Specifically, the child safety seat body 1 provides a safe seat for children in the vehicle. It is a safety-standard compliant seat with appropriate posture support and protection devices, such as headrests and seat belts. This seat can be installed in the rear seat of the vehicle, providing a comfortable riding environment for children. A camera 2 is installed above or to the side of the child safety seat body 1 to monitor the child's posture and status, capturing the child's posture, including sitting upright, reclining, or lying on their side. The captured data is transmitted to a controller 3 for analysis and processing. The controller 3, located inside the child safety seat body, receives image data from the camera and analyzes the child's posture. Based on the analysis results, the controller 3 can make corresponding seat adjustment decisions to ensure the child's posture is safe. More suitable and comfortable; Alarm 4 has an important safety function of the system, used to sound an alarm when abnormal or unsafe child sitting posture is detected; Alarm 4 can include multiple devices, such as car audio, central control display, passenger-side display, rear display, or ceiling display; these devices can emit sound, image, or text alarms to remind the driver and passengers of the status of the child seat; in the system's workflow, camera 2 continuously monitors the child's sitting posture and transmits the data to controller 3 for analysis. Controller 3 detects abnormal sitting postures through image processing algorithms. If an unsafe sitting posture is detected, controller 3 can activate the seat adjustment mechanism. At the same time, if the degree of abnormal sitting posture is high, alarm 4 will be triggered to remind the driver and other people in the vehicle. Combining the functions of the above-mentioned components, the system can monitor the child's sitting posture in real time during vehicle operation and adjust it as needed, thereby improving the safety and comfort of children riding in the car. The addition of alarm 4 also increases the system's timely response capability to abnormal situations.

[0165] In summary, the embodiments of the present invention provide a child safety seat adjustment method, device, and system based on sitting posture detection.

[0166] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0167] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the required tasks. The programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0168] It should also be noted that the exemplary embodiments mentioned in this invention describe methods or systems based on a series of steps or apparatus. However, this invention is not limited to the order of the steps described above; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0169] The above description is merely a specific embodiment of the present invention. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the protection scope of the present invention.

Claims

1. A method for adjusting a child safety seat based on posture detection, characterized in that, The method includes: S1: Acquire a real-time video stream of the child safety seat area and decompose the real-time video stream into multiple real-time images; S2: Perform image processing on each of the real-time images, analyze whether the current sitting posture of the child sitting in the car seat is an abnormal sitting posture, and when it is determined to be an abnormal sitting posture, classify the abnormal sitting posture and output the abnormal sitting posture category. S3: Obtain a preset set of adjustment strategies for abnormal sitting postures of child safety seats, and search for the target strategy corresponding to the abnormal sitting posture category in the set of adjustment strategies according to the abnormal sitting posture category. S4: If the target strategy exists, adjust the child safety seat state according to the target strategy to match the child's sitting posture; S5: If the target strategy does not exist, an alarm message will be issued to remind the user to adjust the child's sitting posture in the car seat; S2 includes: S21: Obtain the preset child's age and the age range information defined by the safety seat; S22: Determine whether the child is using the car seat for the first time, and based on the determination result and the age range information, obtain the standard sitting posture parameters; S23: Obtain the real-time sitting posture parameters of the child in the car seat in the real-time image, compare the standard sitting posture parameters with the real-time sitting posture parameters, and output the abnormal sitting posture category; S22 includes: S221: Obtain a preset usage time interval threshold. If it is determined to be the first use, or not the first use but the usage interval is longer than the threshold, then perform standard sitting posture detection on the current child and output the standard sitting posture parameters. S222: If it is determined that the current child is not using the car seat for the first time and the usage interval is not greater than the duration threshold, then obtain the standard sitting posture parameters of the last use of the car seat.

2. The child safety seat adjustment method based on posture detection according to claim 1, characterized in that, S221 includes: S2211: Input the real-time images into a pre-trained deep learning model to obtain the posture angle information and position information corresponding to the preset parts of the child, wherein the preset parts include at least: head, shoulder, hand and knee. S2212: Analyze and process the attitude angle information and position information, and output the attitude angle variance and position variance; S2213: Compare the attitude angle variance and position variance with preset verification thresholds respectively. When the comparison result meets the verification threshold requirements, output the standard attitude angle information and standard position information at this time as the initial sitting posture parameters. S2214: Obtain real-time feedback information from the user regarding the initial sitting posture parameters. When the user reports that the initial sitting posture parameters meet the requirements, output the current initial sitting posture parameters as the standard sitting posture parameters.

3. The child safety seat adjustment method based on posture detection according to claim 2, characterized in that, S23 includes: S231: Obtain the real-time posture angle information and real-time position information corresponding to the preset parts of the child in the real-time sitting posture parameters; S232: Compare each of the standard attitude angle information with each of the real-time attitude angle information, and compare each of the standard position information with each of the real-time position information, and output the offset parameters corresponding to each preset part; S233: Based on the weight coefficients corresponding to each preset part, perform weighted summation on each offset parameter and output the total offset; S234: Compare the total offset with a preset offset threshold, and output the abnormal sitting posture category based on the comparison result.

4. The child safety seat adjustment method based on posture detection according to claim 3, characterized in that, S234 includes: S2341: Obtain the total target offset corresponding to multiple consecutive frames of images; S2342: Compare the total number of each target offset with the offset threshold, and output the first number of times the total number of target offsets is greater than the offset threshold and the second number of times the total number of target offsets is not greater than the offset threshold; S2343: Output the offset frequency based on the first and second counts; S2344: Compare the offset frequency with a preset frequency threshold and output the abnormal sitting posture category.

5. The child safety seat adjustment method based on posture detection according to claim 4, characterized in that, S4 includes: S41: Obtain the target strategy in the set of adjustment strategies, wherein the target strategy includes at least one of the following strategies: head support height adjustment, seat height adjustment, seat tilt angle adjustment, and shoulder strap position adjustment; S42: Based on the target strategy, control the child safety seat to adjust the sitting posture and obtain the adjusted target sitting posture parameters; S43: Compare the target sitting posture parameters with the standard sitting posture parameters. If the offset frequency corresponding to the target sitting posture parameters is less than the frequency threshold, the adjustment is complete. S44: If the offset frequency corresponding to the target sitting posture parameter is not less than the frequency threshold, a safety reminder is issued to the user.

6. A child safety seat adjustment device based on posture detection, characterized in that, The device includes: The image acquisition module is used to acquire a real-time video stream of the child safety seat area and decompose the real-time video stream into multiple real-time images; The sitting posture detection module is used to process the real-time images and analyze whether the current sitting posture of the child sitting in the car seat is an abnormal sitting posture. When it is determined to be an abnormal sitting posture, the abnormal sitting posture is classified and the abnormal sitting posture category is output. The strategy lookup module is used to obtain a preset set of adjustment strategies for abnormal sitting postures of the child safety seat, and to find the target strategy corresponding to the abnormal sitting posture category in the set of adjustment strategies according to the abnormal sitting posture category. The seating posture adjustment module is used to adjust the state of the child safety seat according to the target strategy if the target strategy exists, so as to match the child's seating posture. The safety warning module is used to issue an alarm message if the target strategy is not present, reminding the user to adjust the child's sitting posture in the car seat; The process involves image processing of each of the real-time images to analyze whether the child's current sitting posture in the car seat is abnormal. When an abnormal posture is determined, it is categorized, and the abnormal posture categories are output, including: Obtain the preset child age and the age range information defined by the car seat; Determine whether the child is using the car seat for the first time, and based on the determination result and the age range information, obtain the standard sitting posture parameters; Obtain the real-time sitting posture parameters of the child in the car seat from the real-time image, compare the standard sitting posture parameters with the real-time sitting posture parameters, and output the abnormal sitting posture category; The determination of whether the child is using the car seat for the first time, based on the determination result and the age range information, includes obtaining standard sitting posture parameters, including: Obtain a preset usage time interval threshold. If it is determined to be the first use, or not the first use but the usage interval is longer than the threshold, then perform standard sitting posture detection on the current child and output the standard sitting posture parameters. If it is determined that the current child is not using the car seat for the first time and the usage interval is not greater than the specified duration threshold, then the standard sitting posture parameters from the last use of the car seat are obtained.

7. A child safety seat adjustment system, the system comprising: Child safety seat body; Camera; Controller; The main body of the child safety seat can be detached and installed inside the vehicle; The camera is used to collect image data and transmit the collected image data to the controller for analysis and processing; The controller receives image data from the camera and analyzes the child's sitting posture; The system is used to implement the child safety seat adjustment method based on posture detection as described in any one of claims 1-5.

8. The child safety seat adjustment system according to claim 7, characterized in that, The system also includes an alarm that receives detection results from the controller, and the alarm includes one or more of the following devices: a car audio system, a central control display screen, a passenger-side display screen, a rear-seat display screen, or a ceiling-mounted display screen.

Images