Intelligent scene adjustment method and device, equipment and storage medium

Abstract

The embodiment of the invention relates to the technical field of information processing, and discloses an intelligent scene adjustment method and device, equipment and a storage medium, and the method comprises the steps: periodically recognizing a user posture appearing in a place; judging whether abnormal postures meeting preset abnormal posture conditions exist in the recognized user postures or not; when it is judged that the abnormal posture exists in the recognized user postures, the scene in the place where the abnormal posture occurs is intelligently adjusted. According to the scene intelligent adjustment method and device, the equipment and the storage medium disclosed by the invention, the problems that a user needs to select a scene and set a triggering condition by himself, use is not convenient enough and scene adjustment cannot be carried out according to an actual use condition in an existing mode are solved, so that the scene in a place is intelligently adjusted based on a user posture, and the user experience is improved. The scene adjustment is performed according to the actual situation without manually selecting the scene or setting the scene adjustment trigger instruction, so that the convenience and the flexibility of scene adjustment performed by the user are effectively improved.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of information processing technology, and in particular to a scene intelligent adjustment method, device, equipment and storage medium. Background Art

[0002] At present, the smart home industry is developing rapidly. It uses residences as carriers and utilizes new-generation information technologies such as the Internet of Things, cloud computing, big data, and artificial intelligence to achieve interconnection between system platforms and home appliances.

[0003] In some technologies, the scenes used in smart homes are all preset default scenes, or scenes created by users. However, when using scenes, users are required to select scenes and set trigger conditions by themselves, which is not convenient to use and cannot adjust scenes according to actual usage. Summary of the invention

[0004] The purpose of the present invention is to at least provide a scene intelligent adjustment method, device, equipment and storage medium, which can at least solve the technical problem that when using the scene, the user needs to select the scene and set the trigger conditions by himself, which is not convenient to use and cannot adjust the scene according to the actual usage situation. At least it can achieve the goal of real-time monitoring of the user posture in the venue, and intelligently adjust the scene in the venue based on the user posture, without the need for manual selection of the scene or setting of the scene adjustment trigger instruction, and the scene adjustment is performed according to the actual usage situation, which effectively improves the convenience and flexibility of the user in adjusting the scene.

[0005] To solve the above technical problems, at least one embodiment of the present application provides a scene intelligent adjustment method, including:

[0006] Periodically identify user gestures occurring in the venue;

[0007] Determining whether there is an abnormal posture that meets a preset abnormal posture condition among the identified user postures;

[0008] When it is determined that the abnormal gesture exists in the recognized user gestures, the scene in the place where the abnormal gesture occurs is intelligently adjusted.

[0009] At least one embodiment of the present application further provides a scene intelligent adjustment device, including:

[0010] An acquisition module, used to periodically identify user gestures appearing in the venue;

[0011] A judging module, used for judging whether there is an abnormal posture that meets a preset abnormal posture condition in the identified user posture;

[0012] The adjustment module is used to intelligently adjust the scene in the place where the abnormal posture occurs when it is determined that the abnormal posture exists in the recognized user posture.

[0013] At least one embodiment of the present application also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned scene intelligent adjustment method.

[0014] At least one embodiment of the present application further provides a computer-readable storage medium storing a computer program, wherein the computer program implements the above-mentioned scene intelligent adjustment method when executed by a processor.

[0015] The embodiments of the present application provide a method for periodically identifying user gestures that appear in a place, and determining whether there are abnormal gestures that meet preset abnormal gesture conditions among the identified user gestures. When it is determined that there are abnormal gestures among the identified user gestures, the scene in the place where the abnormal gesture appears is intelligently adjusted. By real-time monitoring of user gestures in the place, the scene in the place is intelligently adjusted based on the user gestures, without the need to manually select scenes or set scene adjustment trigger instructions, and the scene is adjusted according to actual usage conditions, which effectively improves the convenience and flexibility of users in adjusting scenes.

[0016] In some optional embodiments, the periodic identification of user gestures occurring in the place includes:

[0017] The images in the place that are periodically collected are recognized using a preset human posture recognition model to obtain user postures that appear in the place.

[0018] Through the human posture recognition model, high-quality single-person posture can be extracted, thereby improving the accuracy of posture estimation and achieving better posture estimation results.

[0019] In some optional embodiments, the step of using a preset human posture recognition model to recognize periodically collected images in the place to obtain user postures appearing in the place includes:

[0020] Extracting features of the images in the places in each period respectively to generate feature maps;

[0021] Inputting the feature map into the human body posture recognition model for recognition, and outputting the coordinates of key points used to describe the human body posture;

[0022] Based on the coordinates of the key points, the user posture is determined.

[0023] By detecting each person's bounding box, extracting the features of the image, and inputting the feature map into the human posture recognition model for recognition, the coordinates of each key point are obtained, and then the user posture is determined based on the coordinates of each key point, making the obtained user posture more accurate and the recognition accuracy higher.

[0024] In some optional embodiments, the determining whether there is an abnormal posture that meets a preset abnormal posture condition in the recognized user posture includes:

[0025] A preset abnormal posture classification model is used to classify the user posture output by the human posture recognition model to determine whether there is an abnormal posture that meets the preset abnormal posture condition; wherein the output classification of the abnormal posture detection model includes a normal posture category and an abnormal posture category under at least one of the abnormal posture conditions.

[0026] The user posture is classified through the abnormal posture classification model to determine whether the user posture is abnormal, and the scene of the venue is intelligently adjusted according to the abnormal posture. The scene is adjusted according to the actual usage, which effectively improves the convenience and flexibility of the user in adjusting the scene.

[0027] In some optional embodiments, the venue includes a plurality of sub-areas;

[0028] The determining whether there is an abnormal posture that meets a preset abnormal posture condition in the identified user postures includes:

[0029] For each of the user postures identified in the sub-areas, determine whether there is an abnormal posture that meets the preset abnormal posture condition among the user postures identified in the sub-areas; wherein each of the sub-areas corresponds to a different place function, and the abnormal posture conditions corresponding to the sub-areas with different place functions are also different.

[0030] By judging whether there are abnormal user postures in the sub-areas of the venue, it is possible to intelligently adjust the scenes of the sub-areas separately in the future.

[0031] In some optional embodiments, when it is determined that the recognized user gestures include the abnormal gesture, intelligently adjusting the scene in the place where the abnormal gesture occurs includes:

[0032] When it is determined that the abnormal gesture exists in the user gestures identified in the sub-region, determining the sub-region where the abnormal gesture exists as an adjustment sub-region;

[0033] Determining a scene adjustment task for the adjustment sub-region according to a scene of a sub-region adjacent to the adjustment sub-region;

[0034] Based on the scene adjustment task, the scene of the adjustment sub-area is intelligently adjusted; wherein the scene difference value at the junction between the scene of the adjustment sub-area after the intelligent adjustment and the scene of the sub-area adjacent to the adjustment sub-area is less than a preset difference threshold.

[0035] When the sub-areas in the venue are intelligently adjusted, the transition between the adjusted sub-areas and the adjacent sub-areas is smooth, reducing the abruptness of the connection between the adjusted sub-areas and the adjacent sub-areas.

[0036] In some optional embodiments, the intelligently adjusting the scene in the place where the abnormal gesture occurs includes:

[0037] Intelligently adjust at least one of the temperature, humidity, light color, sound, and light brightness of the scene in the place where the abnormal posture occurs.

[0038] By intelligently adjusting at least one of the temperature, humidity, light color, sound, and light brightness of the smart device, the scene in the venue can be intelligently adjusted. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] One or more embodiments are exemplarily described by the pictures in the corresponding drawings, and these exemplary descriptions do not constitute limitations on the embodiments.

[0040] Figure 1 It is a flowchart of a scene intelligent adjustment method provided by an embodiment of the present application;

[0041] Figure 2 It is a flowchart of a scene intelligent adjustment device provided by an embodiment of the present application;

[0042] Figure 3 is a structural schematic diagram of an electronic device provided by another embodiment of the present application;

[0043] Figure 4 is a flowchart of a scene intelligent adjustment method provided by an embodiment of the present application;

[0044] Figure 5 It is a schematic diagram of sub-area distribution provided by an embodiment of the present application. DETAILED DESCRIPTION

[0045] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the embodiments of the present application will be described in detail below in conjunction with the accompanying drawings. However, it will be appreciated by those skilled in the art that in the present application, many technical details are proposed in order to enable the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical scheme claimed in the present application can also be implemented. The division of the following embodiments is for the convenience of description, and the specific implementation of the present application should not be construed as any limitation, and the various embodiments can be combined and referenced with each other under the premise of no contradiction.

[0046] In order to facilitate understanding of the embodiments of the present application, relevant content about the scene intelligent adjustment method is first introduced here.

[0047] At present, the smart home industry is developing rapidly. It uses residences as carriers and utilizes new-generation information technologies such as the Internet of Things, cloud computing, big data, and artificial intelligence to achieve interconnection between system platforms and home appliances.

[0048] In some technologies, the scenes used in smart homes are all preset default scenes, or scenes created by users. However, when using scenes, users are required to select scenes and set trigger conditions by themselves, which is not convenient to use and cannot adjust scenes according to actual usage.

[0049] In order to solve the technical problem that when using the above-mentioned scenarios, users need to select scenes and set trigger conditions by themselves, which is not convenient to use and cannot adjust the scenes according to actual usage conditions, the present invention proposes a scene intelligent adjustment method. The implementation details of the scene intelligent adjustment method of this embodiment are specifically described below. The following content is only the implementation details provided for the convenience of understanding and is not necessary for the implementation of this solution.

[0050] Embodiment 1:

[0051] The scene intelligent adjustment method of this embodiment can be applied to electronic devices with communication, computing and data storage capabilities. The specific process can be as follows: Figure 1 As shown, including:

[0052] Step 101: periodically identify user gestures in a place.

[0053] Specifically, by setting up monitoring equipment such as cameras in the venue, images of users in the venue are obtained in real time or at intervals of unit time, and the images of users obtained in multiple periods are recognized to obtain the postures of users appearing in the venue.

[0054] Step 102 , determining whether there is an abnormal posture that meets a preset abnormal posture condition among the recognized user postures.

[0055] Specifically, an abnormal posture refers to a posture that is inconsistent with the user posture of the current scene of the venue. When the abnormal posture conditions are different, the abnormal posture is different. In this embodiment, it is determined whether there is an abnormal posture that meets the preset abnormal posture conditions in the identified user postures, in fact, to determine whether the current scene of the venue needs to be adjusted, so as to adjust the scene according to the actual usage.

[0056] Step 103: When it is determined that there is an abnormal posture among the recognized user postures, intelligently adjust the scene in the place where the abnormal posture occurs.

[0057] Specifically, when it is determined that there is an abnormal posture in the recognized user posture, it means that the current scene of the current place needs to be adjusted. In this embodiment, the scene in the place where the abnormal posture occurs is intelligently adjusted. In some examples, each place is equipped with a variety of smart devices, such as smart curtains, smart lamps, smart speakers, smart air conditioners, etc. Each place has a smart scene controlled by smart devices. By adjusting the smart devices in the place where the abnormal posture occurs, the scene in the place is intelligently adjusted.

[0058] In this embodiment, user postures appearing in a place are periodically identified, and it is determined whether there are abnormal postures that meet the preset abnormal posture conditions among the identified user postures. When it is determined that there are abnormal postures among the identified user postures, the scene in the place where the abnormal posture appears is intelligently adjusted. By real-time monitoring of user postures in the place, the scene in the place is intelligently adjusted based on the user posture, without the need to manually select scenes or set scene adjustment trigger instructions, and the scene is adjusted according to the actual usage situation, which effectively improves the convenience and flexibility of users in adjusting scenes.

[0059] In some embodiments, periodically identifying user gestures occurring within a venue includes:

[0060] A preset human posture recognition model is used to recognize images in a venue that are periodically collected to obtain user postures that appear in the venue.

[0061] Specifically, the human posture recognition model is a neural network model, which is implemented using the AlphaPose estimation algorithm based on the regional multi-person posture recognition algorithm (RMPE), and can extract high-quality single-person postures, thereby improving the accuracy of posture estimation, and can obtain good posture estimation results even under inaccurate bounding boxes. In this embodiment, the human posture recognition model is used to recognize images in the venue to obtain user postures appearing in the venue.

[0062] In some embodiments, the human body posture recognition model is obtained by the following steps:

[0063] Constructing the original recognition model;

[0064] The original recognition model is trained based on the acquired user image information samples of multiple periods and the corresponding user posture samples to obtain a human posture recognition model.

[0065] Specifically, by collecting image information samples of users in multiple periods in history and corresponding user posture samples as training sample data, the original recognition model is trained using a reinforcement learning algorithm or a deep learning algorithm to obtain a human posture recognition model.

[0066] In some embodiments, a preset human posture recognition model is used to recognize images in a place that are periodically collected, and the user postures that appear in the place are obtained, including:

[0067] Extract the features of the images in the places of each period respectively and generate a feature map;

[0068] The feature map is input into the human posture recognition model for recognition, and the coordinates of the key points used to describe the human posture are output;

[0069] Based on the coordinates of the key points, the user posture is determined.

[0070] Specifically, after acquiring images in a place for multiple cycles, the physical signs of the images are extracted, which is implemented by using the AlphaPose estimation algorithm based on the regional multi-person posture recognition algorithm (RMPE). Multiple frames are obtained according to the image, and the bounding box of each person is detected. Specifically, slide on the image, extract local features, obtain X, Y, Z corresponding to the color (R, G, B), and compare according to the two pigment blocks adjacent to the pixel point. If the colors are consistent, they are defined as the same color block. A feature map is thus generated to capture different features in the image (such as edges, corners, etc.). Thereafter, the feature map is input into the human posture recognition model for recognition. Since the user's posture has a slope, for example, a person's arm is composed of two joints, not a straight line, the following relationship is used to express the entire posture: posture value = {(input 1*weight 1) + (input 2*weight 2) +...} + bias value, where the posture value is used to reflect the entire posture, the input refers to the slope formula of the posture, and the bias value>0. Then, the coordinates of each key point are output, for example, human body posture 1 = X1, Y1, Z1, X11, Y11, Z11, X12, Y12, Z12, ...; human body posture 2 = X2, Y2, Z2, X3, Y3, Z3, X5, Y5, Z5, X6, Y6, Z6.

[0071] In some examples, multiple user gestures can form a gesture combination, and the gesture combination can reflect user behavior in a certain scene in a certain place.

[0072] In some embodiments, determining whether there is an abnormal gesture that meets a preset abnormal gesture condition among the recognized user gestures includes:

[0073] A preset abnormal posture classification model is used to classify the user posture output by the human posture recognition model to determine whether there is an abnormal posture that meets the preset abnormal posture conditions; wherein the output classification of the abnormal posture detection model includes a normal posture category and an abnormal posture category under at least one abnormal posture condition.

[0074] Specifically, the abnormal posture classification model is a neural network model. After the human posture recognition model outputs the user posture, the abnormal posture classification model classifies the user posture to determine whether the user posture belongs to the normal posture category or the abnormal posture category. When the user posture belongs to the normal posture category, it indicates that no event requiring scene adjustment has occurred in the venue, and there is no need to adjust the scene in the venue; when the user posture belongs to the abnormal posture category, it indicates that an event requiring scene adjustment has occurred in the venue, and the scene in the venue needs to be adjusted. In some examples, the number of abnormal posture conditions can be one or more, and each abnormal posture condition corresponds to an abnormal posture category. For example, when the abnormal posture condition is that there are different sub-areas in the venue and a certain sub-area needs to be adjusted, the abnormal posture category is a partition scene category, and when the abnormal posture condition is that the posture of a user in the venue is different from that of the other users, or the posture of a user is different from the previous posture, the abnormal posture category is a special posture category. For example, the current execution of the cafe scene mode, the current scene is playing music, if the user posture is identified as a special posture category (falling), the page music is stopped.

[0075] In some embodiments, the abnormal posture classification model is obtained by the following steps:

[0076] Build the original classification model;

[0077] The original classification model is trained according to the user posture samples and posture category samples to obtain the abnormal posture classification model.

[0078] Specifically, user posture samples include normal posture samples and abnormal posture samples, and posture category samples include normal posture categories and abnormal posture categories. By collecting user posture samples and posture category samples from multiple periods in history as training sample data, the original classification model is trained using a reinforcement learning algorithm or a deep learning algorithm to obtain an abnormal posture classification model.

[0079] In some embodiments, the venue includes multiple sub-areas;

[0080] Determining whether there is an abnormal posture that meets a preset abnormal posture condition in the recognized user posture includes:

[0081] For the user postures identified in each sub-area, it is determined whether there is an abnormal posture that meets the preset abnormal posture conditions among the user postures identified in each sub-area; wherein each sub-area corresponds to a different place function, and the abnormal posture conditions corresponding to sub-areas with different place functions are also different.

[0082] Specifically, if Figure 5 As shown, the place contains multiple sub-areas, wherein the multiple sub-areas can be closed sub-areas or open sub-areas, and the multiple sub-areas are combined to form a place, such as area A and area B are combined to form a place. In some examples, the sub-area is only a certain area in the place, and the sub-areas cannot be combined to form a place, such as area A cannot be formed to form place B. When abnormal posture recognition is performed on each sub-area separately, it is to determine whether there is a situation in which the user's posture is an abnormal posture in each sub-area. It should be noted that among the multiple sub-areas, there can be sub-areas with the same place function, or there can be sub-areas with different place functions. Accordingly, when the place function of each sub-area is different, the abnormal posture conditions corresponding to the sub-areas with different place functions are also different. In some examples, there are two sub-areas in the place, one sub-area is a lunch break area, and the other sub-area is a study area. The abnormal posture condition in the lunch break area can be that the user's posture is sitting or standing, and the abnormal posture condition in the study area can be that the user's posture is lying or standing. In some examples, if the place functions of all sub-areas in the place are the same, such as the place functions of all sub-areas are used for learning, then the abnormal posture condition is that the user's posture is lying or standing. In some examples, the venue includes area A and area B, and the human body posture in area A of the venue is identified as posture combination 1 (standing, waving hands constantly, and frequently walking in area A); the human body posture in area B of the venue is identified as posture combination 2 (sitting), then the abnormal posture category can be determined as the partition scene category.

[0083] In some embodiments, when it is determined that there is an abnormal gesture in the recognized user gestures, intelligently adjusting the scene in the place where the abnormal gesture occurs includes:

[0084] When it is determined that there is an abnormal posture in the user posture recognized in the sub-region, determining the sub-region where the abnormal posture exists as an adjustment sub-region;

[0085] Determining a scene adjustment task for the adjustment sub-region according to a scene of a sub-region adjacent to the adjustment sub-region;

[0086] Based on the scene adjustment task, the scene of the adjustment sub-region is intelligently adjusted; wherein the scene difference value at the junction between the scene of the adjustment sub-region after the intelligent adjustment and the scene of the sub-region adjacent to the adjustment sub-region is less than a preset difference threshold.

[0087] Specifically, when it is determined that there is an abnormal posture among the user postures recognized in at least one sub-area of ​​multiple sub-areas of the venue, the sub-area with the abnormal posture (i.e., the sub-area) is a sub-area where the scene needs to be intelligently adjusted, and the scene of the adjusted sub-area needs to be intelligently adjusted.

[0088] It should be understood that when intelligently adjusting the scene in a place, not only the function of the place as a whole is taken into account, but also the function of the place divided into different sub-areas. When the functions of the places in different sub-areas are different, simply intelligently adjusting the scene of the adjustment sub-area can easily lead to the transition between the adjustment sub-area and the adjacent sub-area being not harmonious enough, which appears particularly abrupt. For example, a coffee shop has a dining area and a sales area, and the dining area is the adjustment sub-area. If the lighting in the dining area is simply adjusted, it is easy to make the connection between the two open areas appear particularly abrupt. In this embodiment, according to the scene of the sub-area adjacent to the adjustment sub-area, the scene adjustment task of the adjustment sub-area is determined, and then the scene of the adjustment sub-area is intelligently adjusted based on the scene adjustment task. After the scene of the adjustment sub-area is intelligently adjusted, the scene difference value at the connection between the scene of the adjustment sub-area after intelligent adjustment and the scene of the sub-area adjacent to the adjustment sub-area is less than the preset difference threshold. In some examples, a coffee shop has a dining area and a sales area, and the dining area is an adjustment sub-area. The dining area is adjacent to the sales area, and the scene adjustment task of the dining area is determined according to the scene of the sales area (the color temperature of the smart light is 3000K). The scene adjustment task of the dining area is that the color temperature of the smart light is 4500K. The scene of the dining area is intelligently adjusted based on the scene adjustment task, and the color temperature difference of the smart light at the junction between the dining area and the sales area is less than the preset difference threshold.

[0089] In some examples, the venue includes a lunch break area and a study area, both of which are adjustment sub-areas, and the study area is adjacent to the lunch break area. The scene adjustment task of the lunch break area is determined according to the scene of the study area. The scene adjustment task of the lunch break area is to control the closing of the smart curtains in the lunch break area. Based on the scene adjustment task of the lunch break area, the scene of the lunch break area is intelligently adjusted. The scene adjustment task of the study area is determined according to the scene of the lunch break area. The scene adjustment task of the study area is to turn on the desk lamp in the study area. Based on the scene adjustment task of the study area, the scene of the study area is intelligently adjusted, and the brightness difference at the junction of the scene between the study area and the lunch break area is less than the preset difference threshold.

[0090] In some embodiments, intelligently adjusting the scene in a place where an abnormal gesture occurs includes:

[0091] Intelligently adjust at least one of the temperature, humidity, light color, sound, and light brightness of the scene in the place where the abnormal posture occurs.

[0092] Specifically, each place is equipped with a variety of smart devices, such as smart curtains, smart lamps, smart speakers, smart air conditioners, etc. Each place has a smart scene controlled by smart devices. By adjusting the smart devices in the place where abnormal postures occur, at least one of the temperature, humidity, light color, sound, and light brightness of the smart devices can be intelligently adjusted, thereby making intelligent adjustments to the scenes in the place.

[0093] Embodiment 2:

[0094] like Figure 4 As shown, this embodiment provides exemplary content of embodiment one, that is, provides an exemplary process of a scene intelligent adjustment method, and the specific content includes: a smart home management platform, a mobile terminal (central control), and smart devices (including millimeter wave radar).

[0095] S1. Smart building equipment binding:

[0096] (1) Home central control device networking:

[0097] ① Use a mobile terminal (such as App) to scan the QR code on the screen of the home control device (no account is bound, and the QR code data content is fixed) to obtain the ID information of the home control device.

[0098] ② At the same time, turn on the Bluetooth function of the mobile terminal, use Bluetooth to connect to the home control device, and enter the WiFi name or password you want to connect to.

[0099] ③The home host obtains the WiFi name or password information, connects to WiFi, and registers the home central control device ID to the server.

[0100] (2) Add sub-devices to the central control device:

[0101] ① (After completing step 1 above, continue) Use a mobile terminal (such as App) to connect to the discovered WiFi+BLE (GMesh) device through the Gmesh Bluetooth protocol, send WiFi information to the child device, wait for the first device to reply that the router is successfully connected, and wait for the subsequent device to reply that the WiFi information is successfully received. The upper limit of the device connection time is 90s, and the timeout is a connection failure. When the device is connected to the router, if the SSID cannot be searched or the connection to the router fails due to other reasons, it is counted as a connection failure.

[0102] ② For successfully connected devices, the home central control device uploads the home central control device ID and the device ID information of the added sub-device to the server. Set the room location of the device.

[0103] As shown in Table 1 below, the default scene is formed: set the scene, including the use place, the commonly used smart device categories in the place, the type of people (single / multiple), and the scene mode.

[0104] Table 1 Relationship between places and scenes

[0105]

[0106]

[0107] S2. Human posture model training

[0108] (1) The AlphaPose estimation algorithm based on the regional multi-person pose recognition algorithm (RMPE) is used to obtain multiple frames of images and detect the bounding box of each person.

[0109] ① Slide on the image, extract local features, and obtain the X, Y, and Z corresponding to the color. Compare two adjacent color blocks according to the pixels. If the colors are the same, they are defined as the same color block. This generates a feature map to capture different features in the image (such as edges, corners, etc.).

[0110] ② Training image, value = {(input 1*weight 1) + (input 2*weight 2) + ...} + bias value; (bias value > 0), because an action has a slope, like a person's arm, which is composed of two joints, not a straight line, so the above relationship is needed to express the entire posture.

[0111] ③ Output the coordinates of each key point (such as x and y coordinates).

[0112] (2) It can be concluded that human body posture movement 1 = X1, Y1, Z1, X11, Y11, Z11, X12, Y12, Z12...; human body posture movement 2 = X2, Y2, Z2, X3, Y3, Z3, X5, Y5, Z5, X6, Y6, Z6.

[0113] (3) Partition scene model: The human body posture in area A of the venue is identified as combination 1 (standing, waving hands constantly, and frequently walking in area A); the human body posture in area B of the venue is identified as combination 2 (sitting). This is defined as the partition type.

[0114] (4) Special case model: Recognizes human body postures as combination 3 (falling to the ground) and combination 4 (hanging upside down).

[0115] In this embodiment, the model input parameters are a bunch of feature maps of the same type, such as thousands of pictures of the same action, such as sitting, with different sitting positions and image presentation angles, and the model output parameters are user postures.

[0116] S3. Start executing a certain scene, and the cloud sends the device execution instructions. According to the image recognition of the number of people and posture combinations, if the location partition of the person is recognized, the scene adjustment is triggered.

[0117] (1) If the coffee shop scene mode is currently executed, the device is running in Europe, and the recognized human posture combination and position are similar to the training model, the scene adjustment = scene * weight + segmented scene * offset, then the segmented scene is executed, the dining area: the color temperature of the smart light is 4500K, and soft music is played; the sales area: the color temperature of the smart light is 3000K. Among them, the segmented scene is to recognize the posture of the special situation model. That is, when a certain scene is started (such as the coffee shop is open for business, such as B), the human posture of the special situation model appears during the execution process, and the scene is adjusted in this part of the area. (For example, the scene in area A is fine-tuned)

[0118] (2) If the coffee shop scene mode is currently being executed and music is playing; if a special posture (falling) is recognized, the music will stop playing.

[0119] Embodiment three:

[0120] Another embodiment of the present application relates to a scene intelligent adjustment device. The implementation details of the scene intelligent adjustment device of this embodiment are specifically described below. The following content is only for the convenience of understanding the implementation details provided, and is not necessary for the implementation of this solution. The schematic diagram of the scene intelligent adjustment device of this embodiment can be as follows Figure 2 As shown, including:

[0121] An acquisition module 201 is used to periodically identify user gestures that appear in a place;

[0122] The judging module 202 is used to judge whether there is an abnormal posture that meets a preset abnormal posture condition among the recognized user postures;

[0123] The adjustment module 203 is used to intelligently adjust the scene in the place where the abnormal posture occurs when it is determined that there is an abnormal posture in the recognized user posture.

[0124] In some embodiments, the acquisition module 201 is further used to use a preset human posture recognition model to recognize the periodically collected images in the venue to obtain the user postures appearing in the venue.

[0125] In some embodiments, the acquisition module 201 includes:

[0126] An extraction unit, used to extract features of images in the places of each period respectively and generate a feature map;

[0127] A coordinate output unit, used to input the feature map into a human posture recognition model for recognition, and output the coordinates of key points used to describe the human posture;

[0128] The posture determination unit is used to determine the user posture based on the coordinates of the key points.

[0129] In some embodiments, the judgment module 202 is also used to classify the user posture output by the human posture recognition model using a preset abnormal posture classification model to determine whether there is an abnormal posture that meets the preset abnormal posture conditions; wherein the output classification of the abnormal posture detection model includes a normal posture category and an abnormal posture category under at least one abnormal posture condition.

[0130] In some embodiments, the judgment module 202 is also used to judge whether there is an abnormal posture that meets the preset abnormal posture conditions among the user postures recognized in each sub-area; wherein the place includes multiple sub-areas; each sub-area corresponds to a different place function, and the abnormal posture conditions corresponding to sub-areas with different place functions are also different.

[0131] In some embodiments, the adjustment module 203 includes:

[0132] A sub-region determining unit, configured to, when it is determined that an abnormal posture exists in the user posture recognized in the sub-region, determine the sub-region where the abnormal posture exists as an adjustment sub-region;

[0133] A task determination unit, used to determine a scene adjustment task for the adjustment sub-region according to scenes of sub-regions adjacent to the adjustment sub-region;

[0134] An adjustment unit is used to intelligently adjust the scene of the adjustment sub-area based on the scene adjustment task; wherein the scene difference value at the junction between the scene of the adjustment sub-area after the intelligent adjustment and the scene of the sub-area adjacent to the adjustment sub-area is less than a preset difference threshold.

[0135] In some embodiments, the adjustment module 203 is also used to intelligently adjust at least one of the temperature, humidity, light color, sound, and light brightness of the scene in the place where the abnormal posture occurs.

[0136] It is worth mentioning that all modules involved in this embodiment are logic modules. In practical applications, a logic unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, in order to highlight the innovative part of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed by this application, but this does not mean that there are no other units in this embodiment.

[0137] Embodiment 4:

[0138] Another embodiment of the present application relates to an electronic device, such as Figure 3 As shown, it includes: at least one processor 901; and a memory 902 that is communicatively connected to the at least one processor 901; wherein the memory 902 stores instructions that can be executed by the at least one processor 901, and the instructions are executed by the at least one processor 901 so that the at least one processor 901 can execute the scene intelligent adjustment method in the above-mentioned embodiments.

[0139] Among them, the memory and the processor are connected in a bus manner, and the bus may include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. The data processed by the processor is transmitted on a wireless medium via an antenna, and further, the antenna also receives data and transmits the data to the processor.

[0140] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory can be used to store data used by the processor when performing operations.

[0141] Embodiment five:

[0142] Another embodiment of the present application relates to a computer-readable storage medium storing a computer program, which implements the above method embodiment when executed by a processor.

[0143] That is, those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including a number of instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as: ROM), random access memory (Random Access Memory, referred to as: RAM), disk or optical disk and other media that can store program codes.

[0144] Those skilled in the art will appreciate that the above embodiments are specific embodiments for implementing the present application, and in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present application.

Claims

1. A scene intelligent adjustment method, characterized in that: include: Periodically identify user gestures occurring in the venue; Determining whether there is an abnormal posture that meets a preset abnormal posture condition among the identified user postures; When it is determined that the abnormal gesture exists in the recognized user gestures, the scene in the place where the abnormal gesture occurs is intelligently adjusted.

2. The scene intelligent adjustment method according to claim 1, characterized in that: The periodic identification of user gestures occurring in the location includes: The images in the place that are periodically collected are recognized using a preset human posture recognition model to obtain user postures that appear in the place.

3. The scene intelligent adjustment method according to claim 2, characterized in that: The method of using a preset human posture recognition model to recognize the periodically collected images in the place to obtain the user postures appearing in the place includes: Extracting features of the images in the places in each period respectively to generate feature maps; Inputting the feature map into the human body posture recognition model for recognition, and outputting the coordinates of key points used to describe the human body posture; Based on the coordinates of the key points, the user posture is determined.

4. The scene intelligent adjustment method according to claim 2, characterized in that: The determining whether there is an abnormal posture that meets a preset abnormal posture condition in the identified user postures includes: A preset abnormal posture classification model is used to classify the user posture output by the human posture recognition model to determine whether there is an abnormal posture that meets the preset abnormal posture condition; wherein the output classification of the abnormal posture detection model includes a normal posture category and an abnormal posture category under at least one of the abnormal posture conditions.

5. The scene intelligent adjustment method according to claim 4, characterized in that: The venue includes a plurality of sub-areas; The determining whether there is an abnormal posture that meets a preset abnormal posture condition in the identified user postures includes: For each of the user postures identified in the sub-areas, determine whether there is an abnormal posture that meets a preset abnormal posture condition among the user postures identified in the sub-areas; wherein each of the sub-areas corresponds to a different place function, and the abnormal posture conditions corresponding to the sub-areas with different place functions are also different.

6. The scene intelligent adjustment method according to claim 5, characterized in that: When it is determined that the recognized user gestures include the abnormal gesture, intelligently adjusting the scene in the place where the abnormal gesture occurs includes: When it is determined that the abnormal gesture exists in the user gestures identified in the sub-region, determining the sub-region where the abnormal gesture exists as an adjustment sub-region; Determining a scene adjustment task for the adjustment sub-region according to a scene of a sub-region adjacent to the adjustment sub-region; Based on the scene adjustment task, the scene of the adjustment sub-area is intelligently adjusted; wherein the scene difference value at the junction between the scene of the adjustment sub-area after the intelligent adjustment and the scene of the sub-area adjacent to the adjustment sub-area is less than a preset difference threshold.

7. The scene intelligent adjustment method according to claim 1, characterized in that: The intelligently adjusting the scene in the place where the abnormal posture occurs includes: Intelligently adjust at least one of the temperature, humidity, light color, sound, and light brightness of the scene in the place where the abnormal posture occurs.

8. A scene intelligent adjustment device, characterized in that: include: An acquisition module, used to periodically identify user gestures appearing in the venue; A judging module, used for judging whether there is an abnormal posture that meets a preset abnormal posture condition in the identified user posture; The adjustment module is used to intelligently adjust the scene in the place where the abnormal posture occurs when it is determined that the abnormal posture exists in the recognized user posture.

9. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the scene intelligent adjustment method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the scene intelligent adjustment method described in any one of claims 1 to 7 is implemented.

Images