Mattress self-adaption method, device and equipment for realizing deep sleep by adopting AI algorithm

Abstract

The invention relates to the technical field of smart home, solves the technical problems that a mattress cannot be adaptively adjusted in a deep sleep state and the sleep quality of a user is poor in the prior art, and provides a mattress adaptive method, device and equipment for realizing deep sleep by adopting an AI algorithm. The method comprises the steps of processing multiple frames of target images of a target area in a sleep scene through a respiration detection algorithm, and determining respiration frequency information of a target object; judging a real-time sleep state according to the breathing frequency information and the human body sign data; if the sleep state is the deep sleep state, acquiring a real-time sleep posture according to a matching result between the real-time key point information and the standard key point information; according to the real-time sleeping posture, each air bag in the mattress is adjusted to be in a target state conforming to the real-time sleeping posture, and reverse S-shaped supporting of the user is achieved. The problem that the mattress cannot be adaptively adjusted according to the real-time posture of the user in deep sleep in the prior art is effectively solved, and the sleep quality of the user is improved.

Description

[0001] This application is a divisional application of the invention patent application filed on December 29, 2023, with the invention name “Smart mattress control method, device and equipment based on AI algorithm and reverse S support” and application number 202311865193.2. Technical Field

[0002] The present invention relates to the field of smart home technology, and in particular to a mattress self-adaptation method, device and equipment for achieving deep sleep using an AI algorithm. Background Art

[0003] As the quality of life continues to improve, the comfort requirements for beds and chairs are further enhanced, and the concept and market of private customization are also expanding. The starting point of customers for private customization in the field of beds and chairs is not just the superficial requirements of customers who expect to be different. The essence is that they expect the products they buy to change with changes in their height, weight, psychology and living habits.

[0004] Different sleeping states have different effects on sleep quality. Deep sleep is essential for the body's recovery and repair. It is necessary to accurately determine the user's deep sleep state and adjust the mattress hardness, temperature or support accordingly to maximize deep sleep and improve overall sleep quality. Through personalized mattress adaptive adjustment, the mattress can better meet the user's unique needs and provide a more comfortable sleeping experience.

[0005] After determining that the user has entered a deep sleep state, the user will be in various sleeping positions. For example, when sleeping on the bed, the previous habit is to lie on the back, but recently, due to respiratory muscle relaxation and other reasons such as difficulty breathing, the user has changed to lying on the side. If the user is used to lying on the back during sleep, the thickness of the comfort support layer of the mattress for the human body is mainly the support height of the lumbar arch. If it is too much greater than this height, the buttocks will drop too much and the legs will be upturned. It is like sleeping in a round-bottomed pot. In the long run, the spine will inevitably be deformed and cause lumbar convexity. However, when sleeping on the side, if the comfort layer is still the thickness of the supine comfort support layer, the hip bone will touch the hard support layer in the mattress and feel uncomfortable because the thickness of the comfort layer is insufficient, and the waist will also bend downward because the waist side cannot be fully supported by the comfort layer. In the long run, it will also cause scoliosis. Therefore, the immutable structural factors of the existing mattresses are difficult to meet the various changing needs of users, so customers expect to obtain a tailor-made private product that is truly suitable for them. However, the proportions of each of our body structures change over time, and the immutable structure of previous mattresses, even if the products were customized in a certain state, could not meet the real needs and expectations of customers due to their solid-state immutability. The existing Chinese patent CN111053396A discloses a soft and hard adjustable bed chair mattress and its adjustment method, including: a mattress body, which includes a support plate, contact airbags arranged in sequence from the head of the bed to the foot of the bed and capable of sensing whether they are flattened, and a comfort layer laid on the contact airbags; an air pressure device, which includes an air pump for providing pressurized gas to the contact airbags, solenoid valves connected to the contact airbags one by one, a gas pressure detector connected to the interface of each of the solenoid valves, and an exhaust solenoid valve connected to the gas pressure detector; a human physiological detection device, used to detect whether a human body lying on the comfort layer has body movement; and a controller, used to operate the air pressure device to determine the position of different areas of the human body and adjust the pressure of the contact airbags corresponding to different areas of the human body when the human physiological detection device detects body movement, so that the supporting force of the contact airbags on different areas of the human body is evenly distributed. The above patented design has some advantages, such as adjusting the support force of the mattress according to the body movements of different areas of the human body, but it may also have some disadvantages: for example, when there are other non-human objects on the bed, it will affect the contact airbag's judgment of pressure, thereby causing misjudgment and leading to incorrect adjustment of the airbag.

[0006] Therefore, how to improve the user's sleep quality through adaptive adjustment of the mattress in the deep sleep state is an urgent problem to be solved. Summary of the invention

[0007] In view of this, the present invention provides a mattress adaptive method, device and equipment for deep sleep using an AI algorithm, so as to solve the problem in the prior art that the mattress cannot be adaptively adjusted in the deep sleep state, resulting in poor sleep quality for users.

[0008] The technical solution adopted by the present invention is:

[0009] In a first aspect, the present invention provides a mattress adaptation method for achieving deep sleep using an AI algorithm, the method comprising:

[0010] Through a preset breathing detection algorithm, multiple frames of target images of the target area in the sleeping scene are processed to determine the breathing frequency information of the target object;

[0011] According to the breathing frequency information, combined with the human body sign data of the target object, the real-time sleep state of the target object is determined;

[0012] If the real-time sleep state is a deep sleep state, switch to a deep sleep adaptive working mode, wherein the deep sleep adaptive working mode is used to obtain the real-time sleeping posture of the target object according to the matching result between the real-time key point information corresponding to the target object and the standard key point information corresponding to the preset sleeping posture;

[0013] According to the real-time sleeping posture, each airbag in the mattress is adjusted to a target state that matches the real-time sleeping posture, thereby achieving reverse S support for the user.

[0014] Preferably, if the real-time sleeping state is a deep sleeping state, obtaining the real-time sleeping posture of the target object according to the matching result between the real-time key point information corresponding to the target object and the standard key point information corresponding to the preset sleeping posture includes:

[0015] Using a pre-trained deep learning-based skeleton detection algorithm, human body key point detection is performed on each frame of the target image to obtain real-time key point information corresponding to the target object, wherein the real-time key point information includes coordinates of various parts of the body;

[0016] Preset various sleeping positions, including supine, prone and side-lying, and predefine standard key point information for each of the preset sleeping positions, wherein the standard key point information includes preset coordinates for each preset sleeping position;

[0017] Matching the real-time key point information with the standard key point information corresponding to each preset sleeping posture to determine a matching result, wherein the matching result is a similarity score of each preset sleeping posture;

[0018] According to the real-time key point information, pressure data corresponding to each preset part of the target object is obtained;

[0019] The real-time sleeping posture of the target object is determined based on the matching result and the pressure data corresponding to each preset part.

[0020] Preferably, after judging the real-time sleep state of the target object based on the respiratory frequency information and the human body sign data of the target object, the method further includes:

[0021] If the real-time sleep state is an awake state or a preliminary sleeping state, obtaining a preset first adjustment level, a second adjustment level, and a third adjustment level related to the sleep-aiding volume and the massage intensity;

[0022] If the real-time sleeping state is the awake state, adjusting the sleeping aid volume and the massage intensity to the first adjustment level;

[0023] If the real-time sleeping state is a preliminary sleeping state, adjusting the sleeping aid volume and massage intensity to the second adjustment level;

[0024] The time interval after the target object enters the preliminary sleeping state is obtained, and if the time interval is greater than a preset time threshold, the sleeping aid volume and massage intensity are adjusted to the third adjustment level.

[0025] Preferably, the sleep-aid volume and massage intensity corresponding to the first adjustment level are greater than those corresponding to the second adjustment level; the sleep-aid volume and massage intensity corresponding to the second adjustment level are greater than those corresponding to the third adjustment level.

[0026] Preferably, the step of processing multiple frames of target images of a target area in a sleeping scene by a preset breathing detection algorithm to determine the breathing frequency information of the target object includes:

[0027] According to the target images of the consecutive frames, a Gaussian pyramid corresponding to the target images of the consecutive frames is established, wherein the number of levels of each Gaussian pyramid is preset;

[0028] Filtering the corresponding pixel points of each continuous frame of the target image according to the pixel value of each pixel point in each level of the Gaussian pyramid to obtain each target pixel point belonging to weak motion;

[0029] According to each of the target pixel points, corresponding Laplacian pyramids of each layer are established;

[0030] According to each layer of the Laplacian pyramid, the target Laplacian pyramid image is obtained;

[0031] Superimposing the target Laplacian pyramid image with each consecutive frame of the target image to obtain a target video for data enhancement of image data with weak motion;

[0032] According to the video duration of the target video and the number of weak motions, the motion frequency of the weak motion corresponding to the target video is obtained as the breathing frequency information of the target object.

[0033] Preferably, determining the real-time sleeping posture of the target object based on the matching result and in combination with the pressure data corresponding to each preset part includes:

[0034] Acquiring the matching result, wherein the matching result includes: the real-time key point information matches one of the supine sleeping position, the prone sleeping position and the side sleeping position;

[0035] Determine the real-time contact area between each preset part and the mattress according to the pressure data corresponding to each preset part;

[0036] Obtaining a preset area threshold, and if the real-time contact area is greater than the area threshold and the real-time key point information matches the supine sleeping posture or the prone sleeping posture, determining that the real-time sleeping posture of the target object is supine or prone;

[0037] If the real-time contact area is less than or equal to the area threshold and the real-time key point information matches the side-lying sleeping posture, it is determined that the real-time sleeping posture of the target object is side-lying.

[0038] Preferably, adjusting each airbag in the mattress to a target state that matches the real-time sleeping posture according to the real-time sleeping posture to achieve reverse S support for the user includes:

[0039] Determine the target key point positions of the target object according to the real-time key point information, wherein the target key points include: head, neck, left side of waist, center of waist, right side of waist, left hip, right hip, left thigh and right thigh;

[0040] If the real-time sleeping posture is supine or prone, the hardness or height of the airbag in the area where the first target key point is located is adjusted according to the real-time detected position of the first target key point to a target state that matches the real-time sleeping posture, wherein the first target key point includes the center of the head, neck and waist;

[0041] If the real-time sleeping posture is lying on the side, the hardness or height of the airbag in the area where the second target key points are located is adjusted to a target state that matches the real-time sleeping posture, wherein the second target key points include the head, neck, left side of the waist, left buttock and left thigh, or the second target key points include the head, neck, right side of the waist, right buttock and right thigh.

[0042] In a second aspect, the present invention provides a mattress adaptive device for achieving deep sleep using an AI algorithm, the device comprising:

[0043] A breathing detection module is used to process multiple frames of target images of a target area in a sleeping scene through a preset breathing detection algorithm to determine the breathing frequency information of the target object;

[0044] A sleep state determination module, used to determine the real-time sleep state of the target object based on the respiratory frequency information and the target object's physical sign data;

[0045] If the real-time sleep state is a deep sleep state, switch to a deep sleep adaptive working mode, wherein the deep sleep adaptive working mode is used to obtain the real-time sleeping posture of the target object according to the matching result between the real-time key point information corresponding to the target object and the standard key point information corresponding to the preset sleeping posture;

[0046] According to the real-time sleeping posture, each airbag in the mattress is adjusted to a target state that matches the real-time sleeping posture, thereby achieving reverse S support for the user.

[0047] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising: at least one processor, at least one memory, and computer program instructions stored in the memory, which, when executed by the processor, implement the method of the first aspect in the above-mentioned embodiment.

[0048] In a fourth aspect, an embodiment of the present invention further provides a storage medium on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the method of the first aspect in the above-mentioned embodiment is implemented.

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

[0050] The present invention provides a mattress adaptive method, device and equipment for achieving deep sleep using an AI algorithm. The method includes: processing multiple frames of target images of a target area in a sleeping scene through a preset breathing detection algorithm to determine the breathing frequency information of the target object; judging the real-time sleeping state of the target object based on the breathing frequency information and in combination with the human body sign data of the target object; if the real-time sleeping state is a deep sleep state, switching to a deep sleep adaptive working mode, the deep sleep adaptive working mode is used to obtain the real-time sleeping posture of the target object based on the matching result between the real-time key point information corresponding to the target object and the standard key point information corresponding to a preset sleeping posture; based on the real-time sleeping posture, adjusting each airbag in the mattress to a target state that matches the real-time sleeping posture to achieve reverse S support for the user. The present invention monitors the user's breathing rate and vital sign data in real time to accurately identify whether the user is in a deep sleep state. After confirming the deep sleep state, the user's current real-time key point information is further compared with the standard key point information of the preset standard sleeping posture to determine the user's real-time sleeping posture. Through real-time sleeping posture detection, the user's specific posture is identified, thereby providing accurate parameter basis for the adjustment of the mattress; based on the real-time sleeping posture, the inflation state of each airbag in the mattress can be automatically controlled to match the mattress with the user's body posture to achieve an inverted S-shaped support structure. This inverted S-shaped support design fits the natural curve of the human body better, provides good support force and support sense, and then helps the user maintain the best spine and body posture in the deep sleep state, avoiding discomfort and sleep interruption problems caused by improper mattress support. In this way, the present invention effectively solves the problem that the mattress cannot be automatically adjusted according to the user's real-time posture in deep sleep in the prior art, thereby improving the user's sleep quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the drawings required for use in the embodiment of the present invention will be briefly introduced below. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work, and these are all within the protection scope of the present invention.

[0052] Figure 1 This is a schematic diagram of the overall working process of the mattress adaptive method for achieving deep sleep using an AI algorithm in Example 1 of the present invention;

[0053] Figure 2 Schematic diagram of the process of processing each of the target images using a preset breathing detection algorithm in Embodiment 1 of the present invention;

[0054] Figure 3 This is a flow chart of controlling a mattress to switch to a working mode corresponding to a real-time sleeping state in Embodiment 1 of the present invention;

[0055] Figure 4 This is a schematic flow chart of the working steps of the first working mode in Embodiment 1 of the present invention;

[0056] Figure 5 Schematic diagram of the process of the second working mode in Embodiment 1 of the present invention;

[0057] Figure 6 This is a schematic diagram of a process for determining the real-time sleeping posture of the target object in Embodiment 1 of the present invention;

[0058] Figure 7 It is a schematic diagram of a process of adjusting each air bag in the mattress to a target state that matches the real-time sleeping posture in Embodiment 1 of the present invention;

[0059] Figure 8 This is a structural block diagram of a mattress adaptive device for achieving deep sleep using an AI algorithm in Embodiment 2 of the present invention;

[0060] Fig. 9 It is a schematic diagram of the structure of the electronic device in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described in conjunction with the drawings in the embodiment of the present invention. It should be noted that, in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. In the description of the present invention, it should be understood that the orientation or position relationship indicated by the terms "center", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc. is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. Moreover, the term "include", "comprise" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such a process, method, article or device. In the absence of further restrictions, the elements defined by the phrase "comprising..." do not exclude the existence of other identical elements in the process, method, article or device comprising the elements. If there is no conflict, the embodiments of the present invention and the various features in the embodiments can be combined with each other, all within the protection scope of the present invention.

[0062] Example 1

[0063] See also Figure 1 , Embodiment 1 of the present invention discloses a mattress adaptive method for achieving deep sleep using an AI algorithm, wherein the reverse S support means that the mattress can provide intelligent support according to the physical needs of the target object in a supine or prone position, thereby avoiding the user's spine from bending and presenting an S shape, and achieving reverse S support for the user, specifically referring to the use of intelligent technology in the mattress design, by actively matching the user's sleep state and using a sensor unit, an air pressure execution unit combined with an intelligent algorithm, real-time monitoring of the user's sleep state indicators and sleeping posture changes, so as to provide a mattress design with optimal support and pressure distribution, the intelligent mattress comprises: a mattress body, a sensor unit, an execution unit and a control unit; the sensor unit comprises: a pressure sensor, a temperature and humidity sensor and a camera; the execution unit comprises: an airbag adjustment component, a temperature and humidity adjustment component and a music adjustment component; the control unit comprises: a power supply, an MCU, a control circuit and a signal processing component, the method comprises:

[0064] S1: Acquire the human body sign data of the target object and the real-time video data of the target area in the sleeping scene, and decompose the real-time video data into multiple frames of target images;

[0065] Specifically, in the process of obtaining the human body sign data of the target object and the real-time video data of the target area in the sleep scene, first, the user's heart rate and body movement data collected by the physical sign sensor are obtained, wherein the physical sign sensor includes a heart rate monitor and an accelerometer to ensure comprehensive and accurate monitoring of the user's physical sign data; the target objects cover people of all ages, including children, adults and the elderly; since the sleep scene is usually in dark conditions, a thermal infrared camera is used to collect video data, and real-time video data collected by a high-resolution thermal infrared camera is obtained. The thermal infrared camera is installed in a position that can fully cover the target area, such as the ceiling and the head of the bed, to ensure clear and accurate shooting of the target area, and the target area includes the bed surface area, and the real-time video data is decomposed into multiple frames of target images. By comprehensively utilizing the information of the physical sign sensor and the real-time video data, the physiological and behavioral characteristics of the target object during sleep can be fully understood. This comprehensive monitoring method helps to more accurately evaluate sleep quality, detect potential problems, and provide customized sleep support for users of different ages.

[0066] S2: using a preset breathing detection algorithm to process each of the target images to determine the breathing frequency information of the target object;

[0067] In one embodiment, see Figure 2 , said S2 comprises:

[0068] S21: according to the target image of each consecutive frame, establishing a Gaussian pyramid corresponding to the target image of each consecutive frame, wherein the number of levels of each Gaussian pyramid is preset;

[0069] Specifically, data enhancement is performed on each consecutive frame of the target image, specifically, the position offset distance of each pixel point in the image area caused by the respiratory motion is magnified, so as to establish a Laplacian pyramid image corresponding to the image area of ​​the respiratory motion. The Laplacian pyramid image has different levels, and has different spatial frequencies and signal-to-noise ratios. The fewer the levels, the lower the spatial frequency. Preferably, the established Laplacian pyramid image has 3-6 levels.

[0070] S22: Filtering corresponding pixel points of each continuous frame of the target image according to the pixel value of each pixel point in each level of the Gaussian pyramid to obtain each target pixel point belonging to weak motion;

[0071] Specifically, the pixel value of the pixel point in the Gaussian pyramid image is recorded as the first pixel value, and the pixel value of each pixel point in the original image is recorded as the second pixel value; each first pixel value is compared with each second pixel value to obtain a pixel value difference between any first pixel value and each second pixel value, and then each pixel value difference is compared with a pixel value threshold, and the second pixel value corresponding to the pixel value difference less than the pixel value threshold is recorded as a target pixel value, and the pixel point corresponding to the target pixel value is the target pixel point. All target pixel points are screened out by this method, and these target pixel points are not directly subjected to Laplace transformation to establish a Laplace pyramid image, because these target pixel points can be seen to be uniformly changed, so directly performing Laplace transformation will cause holes in the Laplace pyramid image, that is, the gradient of the region will disappear after the Laplace transformation, affecting the accuracy of breathing detection. The input signal is decomposed into wall edge, texture and smooth components. The strong edge refers to the pixels with prominent pixel values ​​that represent the overall contour of the image. The texture refers to the pixels with small differences in pixel values ​​corresponding to the pixels in the strong edge, that is, the pixels that represent the details of the image area corresponding to the pixels in the strong edge. The smooth component is used to enhance the low-frequency components of the image and weaken the high-frequency components to achieve image smoothing. Laplace needs to be derived twice. If the Laplace transform is performed directly, the processing result will be 0 in the color uniform change or gradient area. The processed images in these areas will become holes or disappear, which will affect the accuracy of respiratory monitoring.

[0072] S23: establishing corresponding layers of Laplacian pyramids according to the target pixel points;

[0073] Specifically, after obtaining each target pixel, the Gaussian function is used Each target pixel point is adjusted so that the pixel value of each pixel point in the image with weak motion presents non-uniform change or non-gradual change, so as to avoid the existence of holes or gradient disappearance in the established Laplace pyramid image due to weak motion. f is taken as -2, -1, 2 and 4 respectively, so that the pixel value transformation of the weak motion area presents non-gradual change or uniform change, so as to avoid the pixel value of the area returning to zero after Laplace transform, resulting in the disappearance of the image in the area. After completing one layer of Gaussian pyramid image, a new image is generated, and then the new image is converted into a corresponding layer of Laplace pyramid image. The operation is repeated to finally obtain the target Laplace pyramid image. In this way, the phenomenon of holes or disappearance of the image corresponding to the respiratory motion caused by Laplace change can be avoided, and the integrity of the respiratory motion data can be ensured.

[0074] S24: Obtain the target Laplacian pyramid image according to each layer of the Laplacian pyramid; specifically, obtain each layer of the Laplacian pyramid; use the formula The Laplace pyramid of each layer is magnified and superimposed to obtain the target Laplace pyramid image of the weak movement in the infant sleep video; where I(x, t) is the brightness of the pixel at time t, δ(t) is the displacement distance of the corresponding pixel at time t compared to the previous time, α is the magnification coefficient, and f(x) is the pixel value of pixel x. For each layer of the newly generated local Laplace pyramid, according to the principle of constant brightness of the same pixel at the same time, the following formula is obtained:

[0075] I(x,t)=f(x+δ(t))

[0076] According to the brightness invariance principle, we have:

[0077] I(x,t)=f(x+(1+α)δ(t))

[0078] Considering that sleep breathing movement is a low-frequency movement, low-pass filtering is performed in the low-frequency band, so the first-order Taylor series is expanded as follows:

[0079]

[0080] make:

[0081]

[0082] B(x,t) is the image brightness change signal corresponding to the breathing signal at the spatial point position x and time t in the video channel.

[0083] After magnification and superposition, we get:

[0084]

[0085] By superimposing the enlarged Laplace pyramid image with the original image, we can get pyramid images at different scales. Then, by reconstructing these pyramid images, we can get the enlarged video we need. By amplifying the respiratory movement, it is helpful to increase the recognition speed and accuracy of the respiratory movement, thereby ensuring the accuracy of the detection.

[0086] S25: superimposing the target Laplacian pyramid image with each consecutive frame of the target image to obtain a target video for which data enhancement is performed on image data of weak motion;

[0087] S26: According to the video duration of the target video and the number of weak motions, the motion frequency of the weak motion corresponding to the target video is obtained as the breathing frequency information of the target object.

[0088] Specifically, weak motion includes motion caused by breathing, slight movements of the hands, fluttering of hair, etc. The bands corresponding to the weak motion of the target video are studied to determine the motion frequency corresponding to the weak motion. Specifically, the motion frequency is obtained based on the video length and number of movements of the target video, and the obtained motion frequency is used as the breathing frequency information of the target object.

[0089] S3: judging the real-time sleep state of the target object based on the breathing frequency information and the human body sign data;

[0090] Specifically, based on the respiratory rate information, human body sign data and predefined sleep state classification, the sleep state is classified into four states: awake, asleep, light sleep and deep sleep. These indicators are combined to judge the real-time sleep state of the target object. For example, in the awake state, the respiratory rate is higher, the body movement is active, and the heart rate is higher. The example range is respiratory rate: 15-25 breaths / min, body movement: more frequent exercise, heart rate: 80-100 bpm; in the asleep state, the respiratory rate gradually slows down, the body movement gradually decreases, and the heart rate gradually decreases. The example range is respiratory rate: 10-18 breaths / min, body movement: gradually Reduced, heart rate: 60-80bpm; the respiratory rate and heart rate in the shallow sleep state are relatively stable, with less body movement, example range: respiratory rate: 8-15breaths / min, body movement: relatively still, heart rate: 50-70bpm, the respiratory rate in the deep sleep state is relatively stable, with very little body movement and the lowest heart rate, example range: respiratory rate: 8-12breaths / min, body movement: almost still, heart rate: 40-60bpm, by real-time monitoring of the target object's respiratory rate, body movement and heart rate, comparing this information with the above range, and combining it with the previously defined rules, to determine the target object's current sleep state. This comprehensive judgment helps to more accurately evaluate the sleep state, thereby achieving personalized adjustment of the smart mattress and providing more effective sleep support. It should be noted that these ranges and characteristics are general examples, and specific thresholds and rules need to be optimized according to specific circumstances and individual differences.

[0091] S4: According to the real-time sleeping state of the target object, the mattress is controlled to switch to a working mode corresponding to the real-time sleeping state.

[0092] Specifically, according to the judgment result of the target object's real-time sleep state, the working mode of the mattress is adjusted, and the mattress is controlled to switch to the working mode corresponding to the real-time sleep state. Through such intelligent adjustment, the mattress can better meet the needs of users in different sleep states, provide personalized sleep support, and thus improve the user's sleep experience. At the same time, this personalized adjustment is expected to promote deeper rest and better sleep quality.

[0093] In one embodiment, see Figure 3 , said S4 comprises:

[0094] If the real-time sleep state is the awake state or the preliminary sleeping state, the first working mode of the sleep aid is switched to, wherein the first working mode is used to adjust the sleep aid volume and massage intensity according to a preset adjustment level;

[0095] Specifically, if the real-time sleeping state is determined to be an awake state, a series of soothing functions are implemented by controlling the smart mattress to help the target object better relax the body and fall asleep, including: starting the music playback system and selecting to play sleep-aiding songs in the preset music library, wherein the sleep-aiding songs include soft, soothing music that suits the preferences of the target object and helps relax the brain and body; at the same time, the air bags in the mattress are inflated and deflated to massage the target object's body, and the massage intensity, frequency and location are adjusted to meet the personalized needs of the target object and help the target object fall asleep quickly.

[0096] In one embodiment, see Figure 4 The first working mode is implemented by the following steps:

[0097] Obtaining preset first adjustment level, second adjustment level, and third adjustment level related to sleep-aiding volume and massage intensity;

[0098] Specifically, three pre-set adjustment levels are obtained, which are used to control the sleep-aid volume and massage intensity respectively. Each adjustment level can be set by the user according to personal preference and comfort, or it can be the standard value set by the system by default. For example: the first adjustment level: suitable for the awake state, providing the maximum volume and massage intensity to help the user relax better in the awake state and fall asleep quickly; the second adjustment level: suitable for the initial state of falling asleep, providing a weaker volume and massage intensity to help guide the user into a deeper sleep; the third adjustment level: suitable for a period of time after the initial falling asleep, providing the lightest volume and massage intensity to cope with some challenges that may arise in the process of falling asleep, such as slight awakening.

[0099] If the real-time sleeping state is the awake state, adjusting the sleeping aid volume and the massage intensity to the first adjustment level;

[0100] Specifically, in the awake state, the system will adjust the sleep-aid volume and massage intensity to the pre-set first adjustment level based on the real-time monitored sleep status information. This means that the system will provide the maximum volume and massage intensity to help the target object relax better in the awake state and fall asleep quickly.

[0101] If the real-time sleeping state is a preliminary sleeping state, adjusting the sleeping aid volume and massage intensity to the second adjustment level;

[0102] Specifically, in the initial state of falling asleep, the system will adjust the sleep-aid volume and massage intensity to the preset second adjustment level based on the real-time monitored sleep status information. This level may provide weaker volume and massage intensity, which helps guide the target object into a deeper sleep.

[0103] The time interval after the target object enters the preliminary sleeping state is obtained, and if the time interval is greater than a preset time threshold, the sleeping aid volume and massage intensity are adjusted to the third adjustment level.

[0104] Specifically, the system monitors the time after the target object enters the initial sleep state. If the time interval is greater than the preset time threshold, the system adjusts the sleep-aid volume and massage intensity to the third adjustment level. This level may provide the lightest volume and massage intensity to meet some needs that may arise during the user's sleep, such as slight waking up. Such dynamic adjustments help to personalize the needs of users in different sleep states.

[0105] S42: If the real-time sleep state is a light sleep state, shutting down the first working mode;

[0106] S43: If the real-time sleep state is a deep sleep state, switch to the second working mode of deep sleep adaptation, the second working mode is used to adjust each airbag in the mattress to a target state that matches the real-time sleeping posture according to the real-time sleeping posture of the target object.

[0107] Specifically, if the real-time sleeping state is a deep sleeping state, the airbags in the mattress are adjusted according to the real-time sleeping posture of the target object. By adjusting to the target state that matches the real-time sleeping posture, the mattress can better adapt to the different sleeping postures of the user in the deep sleeping state and provide more comfortable and physiologically consistent support. This helps to improve the user's sleep quality and overall sleeping experience.

[0108] In one embodiment, see Figure 5 The second working mode is implemented by the following steps:

[0109] Using a pre-trained deep learning-based skeleton detection algorithm, human key point detection is performed on the target image of each frame to obtain real-time key point information corresponding to the target object;

[0110] Specifically, a pre-trained skeleton detection algorithm based on deep learning is used to detect human key points of the target image in each frame. The skeleton detection algorithm is completed by a pre-trained skeleton detection algorithm based on deep learning. Specifically, a VGG pre-train network is used as a backbone network, which can extract the features of the target image in each input frame. The VGG pre-train network is used as a backbone network, and the features of the target image in each input frame are extracted through operations such as convolution and pooling. These features play a key role in subsequent skeleton detection; two branch networks are constructed, one of which is used to regress the position of the key point (L(p)), and the other branch is used to generate PartAffinity Fields (PAFs, limb relationship), these two branches are responsible for capturing the key point position and limb relationship of the target object respectively. In each training state, the network adjusts the parameters by calculating the loss function to adapt to the annotated key point position and limb relationship. The loss function includes the L2 norm loss of L(p) and S(p), where L(p) represents the key point position and S(p) represents the limb relationship. This training process is performed in each state until the deep learning network converges; after the training is completed, the deep learning network can generate the predicted position of each key point in the image and the prediction of the limb relationship, which is gradually optimized and generated during each state iteration. PAFs are used to represent the direction and strength of the limb. The deep learning network constructs these Fields by finding the vector direction between the pixels connecting each pair of key points. By matching adjacent nodes, such as matching the left wrist node with the left elbow node, the PAF of the forearm is used to calculate the edge weight, and finally the human posture information of the entire target object is obtained, and the skeleton structure of the human body is established, and the connection and relative position between the key points are identified as the real-time key point information corresponding to the target object.

[0111] Matching the real-time key point information with the standard key point information corresponding to each preset sleeping posture, and outputting the matching result;

[0112] Specifically, real-time human key point information is obtained from the skeleton detection algorithm. These key points include the coordinates of various parts of the body, such as the head, shoulders, wrists, knees, etc.

[0113] Preset various sleeping positions, including supine, prone and side-lying, and predefine standard key point positions for each of the preset sleeping positions, wherein the standard key point information includes preset coordinates for each preset sleeping position;

[0114] Comparing the real-time key point information with the standard key point information corresponding to each preset sleeping posture by calculating the distance or similarity between the real-time key point coordinates and the preset standard coordinates;

[0115] Determine the current sleeping position of the target object according to the matching result, where the matching result may be a similarity score of each preset sleeping position;

[0116] Setting a similarity threshold to determine when the real-time key point information is considered to be successfully matched with a preset sleeping posture;

[0117] If the similarity is higher than the set threshold, the match is considered successful and the one with the highest similarity is selected as the current real-time sleeping posture.

[0118] By matching the real-time key point information with the standard key point information corresponding to the preset sleeping position, the system can identify the target object's sleeping position in real time, providing important information for subsequent sleep state determination and mattress adjustment.

[0119] According to the real-time key point information, pressure data corresponding to each preset part of the target object is obtained;

[0120] Specifically, an array of pressure sensors is arranged on the mattress in advance, and these sensors can sense the pressure distribution on the mattress surface. Each sensor is responsible for monitoring the pressure changes in a specific area; for each preset part, a correspondence between the key point and the corresponding sensor area on the mattress is established. For example, the key point "head" may correspond to the head area on the mattress, the key point "shoulder" corresponds to the shoulder area on the mattress, and so on; using the preset correspondence, the real-time key point information is mapped to the corresponding sensor area on the mattress, so that the specific location of each key point can be determined, and thus associated with the corresponding pressure sensor. Through the pressure sensor on the mattress, the pressure data of each preset part corresponding to the real-time key point information is obtained. These data reflect the pressure distribution of different parts of the target object on the bed.

[0121] According to the matching result, combined with the pressure data corresponding to each preset part, the real-time sleeping posture of the target object is determined, wherein the real-time sleeping posture includes supine, side-lying and prone;

[0122] Specifically, according to the matching result and in combination with the pressure data corresponding to each preset part, the real-time sleeping position of the target object is determined, wherein the real-time sleeping position includes supine, side-lying and prone;

[0123] Combining the real-time key point information and pressure data in the target image, the real-time sleeping position of the target object can be determined more accurately;

[0124] Pressure data can provide the force distribution on the mattress, while real-time key point information provides the specific location of body parts. The combination of the two can more accurately identify sleeping positions and avoid misjudgments that may exist when relying solely on pressure data: for example, there may be other items on the mattress, such as pillows, toys, etc., which will also exert a certain amount of pressure on the mattress. Judging solely by pressure data may be interfered by these additional factors and lead to misjudgments. By combining image information, the body parts of the target object can be identified to avoid misjudgments caused by other items on the mattress. The comprehensive determination of real-time sleeping positions not only helps with personalized mattress adjustments, but also provides more comprehensive sleep analysis, understands the user's sleeping patterns and habits in different sleeping positions, and helps provide users with more personalized sleep suggestions and improvement plans.

[0125] In one embodiment, see Figure 6 , the step of determining the real-time sleeping posture of the target object based on the matching result and the pressure data corresponding to each preset part includes:

[0126] Acquiring the matching result, wherein the matching result includes: the real-time key point information matches one of the supine sleeping position, the prone sleeping position and the side sleeping position;

[0127] Determine the real-time contact area between each preset part and the mattress according to the pressure data corresponding to each preset part;

[0128] Obtaining a preset area threshold, and if the real-time contact area is greater than the area threshold and the real-time key point information matches the supine sleeping posture or the prone sleeping posture, determining that the real-time sleeping posture of the target object is supine or prone;

[0129] If the real-time contact area is less than or equal to the area threshold and the real-time key point information matches the side-lying sleeping posture, it is determined that the real-time sleeping posture of the target object is side-lying.

[0130] According to the real-time sleeping posture of the target object, each airbag in the mattress is adjusted to a target state that matches the real-time sleeping posture.

[0131] In one embodiment, see Figure 7 The step of adjusting each airbag in the mattress to a target state that matches the real-time sleeping posture of the target object includes:

[0132] Determine the target key point positions of the target object according to the real-time key point information, wherein the target key points include: head, neck, left side of waist, center of waist, right side of waist, left hip, right hip, left thigh and right thigh;

[0133] If the real-time sleeping posture is supine or prone, the airbag in the area where the first target key point is located is adaptively adjusted to a target state that matches the real-time sleeping posture, wherein the first target key point includes the center of the head, neck and waist;

[0134] Specifically, if the real-time sleeping posture is supine or prone, the airbags on the mattress are adaptively adjusted for the head area so that the head can obtain appropriate support, which helps to maintain the natural posture of the head and reduce discomfort in the neck and spine; the neck is a key area that needs good support during sleep. By adjusting the corresponding airbags, it is ensured that the neck obtains appropriate support to maintain the normal physiological curve of the cervical spine and reduce neck pressure; for the central area of ​​the waist, the system will adjust the airbags on the mattress to provide support for the lumbar spine, which helps to maintain the natural curvature of the lumbar spine, reduce waist pressure, and thus improve overall sleep comfort; adaptive adjustment refers to dynamically adjusting the hardness or height of the airbag according to the real-time detected target key point position. Through such adaptive adjustment, the mattress can provide intelligent support according to the physical needs of the target object in the supine or prone position, thereby avoiding the user's spinal curve from bending and presenting an S shape, and realizing anti-S support for the user. Anti-S support refers to the use of intelligent technology in the mattress design, by actively matching the user's sleep state and using the sensor unit, the air pressure execution unit and the intelligent algorithm to monitor the user's sleep state indicators and sleeping posture changes in real time to provide the best support and pressure distribution mattress design. This personalized adjustment helps provide a more comfortable and human-friendly sleeping experience, thereby improving the user's sleep onset and sleep quality.

[0135] If the real-time sleeping posture is lying on the side, the airbag in the area of ​​the second target key point where you are located is adaptively adjusted to a target state that matches the real-time sleeping posture, wherein the second target key points include the head, neck, left side of the waist, left buttock and left thigh, or the second target key points include the head, neck, right side of the waist, right buttock and right thigh.

[0136] Specifically, if the real-time sleeping posture is side-lying, the key point positions of the head, neck, left side of the waist, left buttock and left thigh in the side-lying posture are mapped to the airbags in the corresponding area, or the key point positions of the head, neck, right side of the waist, right buttock and right thigh in the side-lying posture are mapped to the airbags in the corresponding area, and a feedback loop adjustment algorithm, i.e., a feedback mechanism, is used to adjust the mattress in real time to provide more comfortable support by sensing the user's feedback or other physiological data. The hardness or height of the airbag in the corresponding area that needs to be adjusted is calculated through the real-time key point information and the mapping relationship, which may include the inflation or deflation of the airbag to provide support that matches the side-lying posture. The adjustment is dynamic, and the system can monitor the changes in the position of the user's head, neck, left side of the waist, left buttock and left thigh, or the head, neck, right side of the waist, right buttock and right thigh in the side-lying posture in real time, and make corresponding airbag adjustments when necessary. Adaptive adjustment allows the system to better adapt to the individual differences and habits of users, provide personalized support, and ensure that users can get the best sleeping experience in the side-lying posture. Through such an adaptive adjustment mechanism, the mattress system can provide more personalized and human physiological support in the side-lying position, helping users fall asleep better and improve overall sleep quality.

[0137] Example 2

[0138] See also Figure 8 Embodiment 2 of the present invention further provides a mattress adaptive device for achieving deep sleep using an AI algorithm, the device comprising:

[0139] A data acquisition module, used to acquire the human body sign data of the target object and the real-time video data of the target area in the sleeping scene, and decompose the real-time video data into multiple frames of target images;

[0140] A breathing detection module, used to process each of the target images using a preset breathing detection algorithm to determine the breathing frequency information of the target object;

[0141] A sleep state analysis module, used to determine the real-time sleep state of the target object based on the respiratory frequency information and the human body sign data;

[0142] The mode switching module is used to control the mattress to switch to a working mode corresponding to the real-time sleeping state according to the real-time sleeping state of the target object.

[0143] Specifically, a mattress adaptive device for achieving deep sleep using an AI algorithm provided by an embodiment of the present invention includes: a data acquisition module, used to acquire human body sign data of a target object and real-time video data of a target area in a sleeping scene, and decompose the real-time video data into multiple frames of target images; a breathing detection module, used to process each of the target images using a preset breathing detection algorithm to determine the breathing frequency information of the target object; a sleep state analysis module, used to judge the real-time sleep state of the target object based on the breathing frequency information and the human body sign data; a mode switching module, used to control the mattress to switch to a working mode corresponding to the real-time sleep state according to the real-time sleep state of the target object. The device obtains human vital signs data and real-time video data, decomposes the real-time video data into multiple frames of target images, and combines the respiratory rate information and human vital signs data to make full use of multiple information sources to judge the user's real-time sleep state, which helps to accurately determine whether the user is in a specific state such as difficulty falling asleep or difficulty falling asleep deeply, and provides strong support for formulating corresponding mattress working modes; at the same time, according to the real-time sleep state of the target object, the solution adjusts the mattress to the corresponding working mode. For example, when it is detected that the user is in an awake state, it is adjusted to provide gentle massage or mild vibration to help the user relax the body and promote falling asleep; when it is detected that the user is in a deep sleep state, the mattress can be adjusted to provide more stable support and adjust the hardness to ensure that the user gets better comfort in deep sleep. This personalized intelligent adjustment helps to improve the user's sleep quality and meet the needs of users in different sleep states, thereby solving the problem of difficulty falling asleep and difficulty falling asleep deeply.

[0144] Example 3

[0145] In addition, combined Figure 1 The mattress adaptation method for achieving deep sleep using an AI algorithm described in Embodiment 1 of the present invention may be implemented by an electronic device. Fig. 9 A schematic diagram of the hardware structure of an electronic device provided in Embodiment 3 of the present invention is shown.

[0146] The electronic device may include a processor and a memory storing computer program instructions.

[0147] Specifically, the processor may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present invention.

[0148] The memory may include a large capacity memory for data or instructions. By way of example and not limitation, the memory may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. Where appropriate, the memory may include a removable or non-removable (or fixed) medium. Where appropriate, the memory may be inside or outside a data processing device. In a particular embodiment, the memory is a non-volatile solid-state memory. In a particular embodiment, the memory includes a read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM) or a flash memory or a combination of two or more of these.

[0149] The processor reads and executes computer program instructions stored in the memory to implement any one of the mattress adaptation methods for achieving deep sleep using an AI algorithm in the above embodiments.

[0150] In one example, the electronic device may further include a communication interface and a bus. Fig. 9 As shown, the processor, memory, and communication interface are connected via a bus and communicate with each other.

[0151] The communication interface is mainly used to implement communication between the modules, devices, units and / or equipment in the embodiments of the present invention.

[0152] The bus includes hardware, software or both, and the parts of the device are coupled to each other. For example, but not limitation, the bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In suitable cases, the bus may include one or more buses. Although the embodiment of the present invention describes and shows a specific bus, the present invention considers any suitable bus or interconnection.

[0153] Example 4

[0154] In addition, in combination with the mattress adaptive method for achieving deep sleep using an AI algorithm in the above embodiment 1, the present invention can also provide a computer-readable storage medium for implementation in embodiment 4. The computer-readable storage medium stores computer program instructions; when the computer program instructions are executed by the processor, any one of the mattress adaptive methods for achieving deep sleep using an AI algorithm in the above embodiments is implemented.

[0155] In summary, the embodiments of the present invention provide a mattress adaptation method, device and apparatus for achieving deep sleep using an AI algorithm.

[0156] It should be clear that the present invention is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is 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, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present invention.

[0157] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present invention are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal 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, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0158] It should also be noted that the exemplary embodiments mentioned in the present invention describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in a different order from the embodiments, or several steps can be performed simultaneously.

[0159] The above is only a specific implementation of the present invention. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the system, module and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be covered within the protection scope of the present invention.

Claims

1. A mattress adaptive method for achieving deep sleep using an AI algorithm, characterized in that: The method comprises: Through a preset breathing detection algorithm, multiple frames of target images of the target area in the sleeping scene are processed to determine the breathing frequency information of the target object; According to the breathing frequency information, combined with the human body sign data of the target object, the real-time sleep state of the target object is determined; If the real-time sleep state is a deep sleep state, switch to a deep sleep adaptive working mode, wherein the deep sleep adaptive working mode is used to obtain the real-time sleeping posture of the target object according to the matching result between the real-time key point information corresponding to the target object and the standard key point information corresponding to the preset sleeping posture; According to the real-time sleeping posture, each airbag in the mattress is adjusted to a target state that matches the real-time sleeping posture, thereby achieving reverse S support for the user.

2. The mattress adaptive method for achieving deep sleep using an AI algorithm according to claim 1, characterized in that: If the real-time sleep state is a deep sleep state, switching to a deep sleep adaptive working mode, the deep sleep adaptive working mode is used to obtain the real-time sleeping posture of the target object according to the matching result between the real-time key point information corresponding to the target object and the standard key point information corresponding to the preset sleeping posture, including: Using a pre-trained deep learning-based skeleton detection algorithm, human body key point detection is performed on each frame of the target image to obtain real-time key point information corresponding to the target object, wherein the real-time key point information includes coordinates of various parts of the body; Preset various sleeping positions, including supine, prone and side-lying, and predefine standard key point information for each of the preset sleeping positions, wherein the standard key point information includes preset coordinates for each preset sleeping position; Matching the real-time key point information with the standard key point information corresponding to each preset sleeping posture to determine a matching result, wherein the matching result is a similarity score of each preset sleeping posture; According to the real-time key point information, pressure data corresponding to each preset part of the target object is obtained; The real-time sleeping posture of the target object is determined based on the matching result and the pressure data corresponding to each preset part.

3. The mattress adaptive method for achieving deep sleep using an AI algorithm according to claim 1, characterized in that: After judging the real-time sleep state of the target object based on the breathing frequency information and the human body sign data of the target object, the method further includes: If the real-time sleep state is an awake state or a preliminary sleeping state, obtaining a preset first adjustment level, a second adjustment level, and a third adjustment level related to the sleep-aiding volume and the massage intensity; If the real-time sleeping state is the awake state, adjusting the sleeping aid volume and the massage intensity to the first adjustment level; If the real-time sleeping state is a preliminary sleeping state, adjusting the sleeping aid volume and massage intensity to the second adjustment level; The time interval after the target object enters the preliminary sleeping state is obtained, and if the time interval is greater than a preset time threshold, the sleeping aid volume and massage intensity are adjusted to the third adjustment level.

4. The mattress adaptive method for achieving deep sleep using an AI algorithm according to claim 3, characterized in that: The sleep-aid volume and massage intensity corresponding to the first adjustment level are greater than the sleep-aid volume and massage intensity corresponding to the second adjustment level; the sleep-aid volume and massage intensity corresponding to the second adjustment level are greater than the sleep-aid volume and massage intensity corresponding to the third adjustment level.

5. The mattress adaptive method for achieving deep sleep using an AI algorithm according to claim 3, characterized in that: The method of processing multiple frames of target images of the target area in the sleep scene by using a preset breathing detection algorithm to determine the breathing frequency information of the target object includes: According to the target images of the consecutive frames, a Gaussian pyramid corresponding to the target images of the consecutive frames is established, wherein the number of levels of each Gaussian pyramid is preset; Filtering the corresponding pixel points of each continuous frame of the target image according to the pixel value of each pixel point in each level of the Gaussian pyramid to obtain each target pixel point belonging to weak motion; According to each of the target pixel points, corresponding Laplacian pyramids of each layer are established; According to each layer of the Laplacian pyramid, the target Laplacian pyramid image is obtained; Superimposing the target Laplacian pyramid image with each consecutive frame of the target image to obtain a target video for data enhancement of image data with weak motion; According to the video duration of the target video and the number of weak motions, the motion frequency of the weak motion corresponding to the target video is obtained as the breathing frequency information of the target object.

6. The mattress adaptive method for achieving deep sleep using an AI algorithm according to claim 2, characterized in that: Determining the real-time sleeping posture of the target object based on the matching result and combining the pressure data corresponding to each preset part includes: Acquiring the matching result, wherein the matching result includes: the real-time key point information matches one of the supine sleeping position, the prone sleeping position and the side sleeping position; Determine the real-time contact area between each preset part and the mattress according to the pressure data corresponding to each preset part; Obtaining a preset area threshold, and if the real-time contact area is greater than the area threshold and the real-time key point information matches the supine sleeping posture or the prone sleeping posture, determining that the real-time sleeping posture of the target object is supine or prone; If the real-time contact area is less than or equal to the area threshold and the real-time key point information matches the side-lying sleeping posture, it is determined that the real-time sleeping posture of the target object is side-lying.

7. The mattress adaptive method for achieving deep sleep using an AI algorithm according to claim 2, characterized in that: According to the real-time sleeping posture, adjusting each airbag in the mattress to a target state that matches the real-time sleeping posture to achieve reverse S support for the user includes: Determine the target key point positions of the target object according to the real-time key point information, wherein the target key points include: head, neck, left side of waist, center of waist, right side of waist, left hip, right hip, left thigh and right thigh; If the real-time sleeping posture is supine or prone, the hardness or height of the airbag in the area where the first target key point is located is adjusted according to the real-time detected position of the first target key point to a target state that matches the real-time sleeping posture, wherein the first target key point includes the center of the head, neck and waist; If the real-time sleeping posture is lying on the side, the hardness or height of the airbag in the area where the second target key points are located is adjusted to a target state that matches the real-time sleeping posture, wherein the second target key points include the head, neck, left side of the waist, left buttock and left thigh, or the second target key points include the head, neck, right side of the waist, right buttock and right thigh.

8. A mattress adaptive device for achieving deep sleep using an AI algorithm, characterized in that: The device comprises: A breathing detection module is used to process multiple frames of target images of a target area in a sleeping scene through a preset breathing detection algorithm to determine the breathing frequency information of the target object; A sleep state determination module, used to determine the real-time sleep state of the target object based on the respiratory frequency information and the target object's physical sign data; If the real-time sleep state is a deep sleep state, switch to a deep sleep adaptive working mode, wherein the deep sleep adaptive working mode is used to obtain the real-time sleeping posture of the target object according to the matching result between the real-time key point information corresponding to the target object and the standard key point information corresponding to the preset sleeping posture; According to the real-time sleeping posture, each airbag in the mattress is adjusted to a target state that matches the real-time sleeping posture, thereby achieving reverse S support for the user.

9. An electronic device, characterized in that: include: At least one processor, at least one memory and computer program instructions stored in the memory, when the computer program instructions are executed by the processor, implement the method according to any one of claims 1 to 7.

10. A storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images