Water replenishing prompting method and device, electronic equipment and storage medium
By predicting the water loss data of human body and determining the hydration prompt strategy, the problem that existing smart devices cannot accurately prompt users to replenish water is solved, the timing and amount of water is improved, and the user's hydration experience is improved.
Patent Information
- Application Number
- CN202411193785.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-28
- Publication Date
- 2024-12-06
AI Technical Summary
Existing smart devices cannot accurately know the timing and amount of water needed for the human body to replenish moisture, resulting in the inability to effectively prompt users to replenish moisture.
By obtaining human model data, environmental data, state intensity data and body surface temperature data, the prediction model is used to predict human body water loss data, and a water replenishment prompt strategy is determined based on this, and sent to the user to prompt for water replenishment.
It accurately prompts users to rehydrate in specific conditions, improving the hydration experience and effect.
Smart Images

Figure CN119073933A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of smart devices, and in particular to a water replenishment prompting method, device, electronic device and storage medium. Background Art
[0002] People know that drinking water regularly is good for health, but they still forget to replenish water when they are in a certain state, such as exercising or working, which causes dehydration and various diseases. There are some smart devices that remind users to drink water, but these smart devices generally send reminders at regular intervals to remind users to drink water. They cannot accurately know when the human body needs to replenish water, nor can they accurately calculate the amount of water required. Therefore, a solution is urgently needed to accurately remind users to replenish water. Summary of the invention
[0003] The embodiment of the present invention provides a hydration reminder method, aiming to provide a hydration reminder solution to accurately remind users to hydrate. When the user is in a specific state, the multi-dimensional data of the user can be used to predict the user's body water loss data, and based on the body water loss data, a hydration reminder strategy for the user can be determined, so that the target user can be reminded to hydrate in a timely and accurate manner, thereby improving the user's hydration experience.
[0004] In a first aspect, an embodiment of the present invention provides a water replenishment reminder method, the method comprising:
[0005] When the target user is in a specific state, multi-dimensional data of the target user is obtained, wherein the multi-dimensional data includes human body model data, environmental data, state intensity data, and body surface temperature data;
[0006] Based on the trained prediction model, the multi-dimensional data is predicted and processed to obtain the human body water loss data corresponding to the target user;
[0007] Based on the human body water loss data, a water replenishment prompt strategy for the target user is determined, and the water replenishment prompt strategy is sent to the target user to prompt the user to replenish water according to the water replenishment prompt strategy.
[0008] Optionally, the prediction model includes a human surface area feature extraction network, an environmental feature extraction network, a state intensity feature extraction network, a sweat evaporation feature extraction network, a feature fusion network, and an output network. The step of performing prediction processing on the multi-dimensional data based on the trained prediction model to obtain the human body water loss data corresponding to the target user includes:
[0009] Performing feature extraction processing on the human body model data through the human body surface area feature extraction network to obtain human body surface features corresponding to the target user;
[0010] Performing feature extraction processing on the environmental data through the environmental feature extraction network to obtain environmental features corresponding to the target user;
[0011] Performing feature extraction processing on the state strength data through the state strength feature extraction network to obtain the state strength feature corresponding to the target user;
[0012] Performing feature extraction processing on the body surface temperature data through the sweat evaporation feature extraction network to obtain sweat evaporation features corresponding to the target user;
[0013] The human body surface features, the environmental features, the state intensity features, and the sweat evaporation amount features are fused through the fusion network to obtain fusion features of the target user;
[0014] The fused features are subjected to linear regression processing through the output network to output the human body water loss data corresponding to the target user.
[0015] Optionally, before the step of performing feature extraction processing on the human body model data through the human body surface area feature extraction network to obtain the human body surface features corresponding to the target user, the method further includes:
[0016] Acquire first training sample data and a three-dimensional human body generation model, wherein the first training sample data includes first sample human body model data and a three-dimensional human body label, and one sample human body model data corresponds to one three-dimensional human body label;
[0017] Training the three-dimensional human body generation model using the first training sample data to obtain a trained three-dimensional human body generation model;
[0018] Based on the trained three-dimensional human body generation model, the human body surface area feature extraction network is generated.
[0019] Optionally, the step of generating the human body surface area feature extraction network based on the trained three-dimensional human body generation model includes:
[0020] Generate a first three-dimensional human body model corresponding to the first sample human body model data by using the trained three-dimensional human body generation model, wherein each first sample human body model data corresponds to one first three-dimensional human body model;
[0021] Performing surface segmentation processing on the first three-dimensional human body model to obtain a first human body surface area feature corresponding to the first three-dimensional human body model;
[0022] Constructing a mapping network between the first sample human body model data and the first human body surface area feature;
[0023] Based on the mapping network, the human body surface area feature extraction network is generated.
[0024] Optionally, the step of generating the human body surface area feature extraction network based on the trained three-dimensional human body generation model includes:
[0025] Acquire second training sample data and a human body surface segmentation model, wherein the second training sample data includes second sample human body model data, a second three-dimensional human body model, a human body surface sweat gland distribution map, and a human body surface sweating area label, and the second three-dimensional human body model is generated by the trained three-dimensional human body generation model and processed by the second sample human body model data;
[0026] Training the human body surface segmentation model based on the second training sample data to obtain a trained human body surface segmentation model;
[0027] Based on the trained human body surface segmentation model, the human body surface area feature extraction network is constructed.
[0028] Optional,
[0029] The environmental data includes wind data, temperature data and humidity data, and the step of performing feature extraction processing on the environmental data through the environmental feature extraction network to obtain the environmental features corresponding to the target user includes:
[0030] Performing feature extraction processing on the wind data, the temperature data, and the humidity data through the environmental feature extraction network to obtain environmental features corresponding to the target user;
[0031] The step of performing feature extraction processing on the state strength data through the state strength feature extraction network to obtain the state strength feature corresponding to the target user includes:
[0032] Determine a state intensity value corresponding to the target user based on the state intensity data and the reference state data;
[0033] Performing feature extraction processing on the state strength value through the state strength feature extraction network to obtain the state strength feature corresponding to the target user;
[0034] The body surface temperature data includes skin temperature data and ambient temperature data, and the step of performing feature extraction processing on the body surface temperature data through the sweat evaporation feature extraction network to obtain the sweat evaporation feature corresponding to the target user includes:
[0035] Determining a skin temperature difference of the target user based on a difference between the skin temperature data and the ambient temperature data;
[0036] The sweat evaporation feature extraction network performs feature extraction processing on the skin temperature difference to obtain the sweat evaporation feature corresponding to the target user.
[0037] Optionally, the step of determining a water replenishment prompt strategy for the target user based on the human body water loss data includes:
[0038] Determining the body water loss amount of the target user based on the body water loss data;
[0039] Determining a total water replenishment amount for the target user based on the water loss of the human body;
[0040] Determining the water replenishment amount of the target user at each water replenishment based on the water replenishment habit of the target user and the total water replenishment amount, wherein the water replenishment habit includes the water replenishment frequency and the water replenishment type;
[0041] Based on the amount of water replenished by the target user during each water replenishment, a water replenishment prompt strategy for the target user is determined.
[0042] In a second aspect, an embodiment of the present invention further provides a water replenishment reminder device, the water replenishment reminder device comprising:
[0043] An acquisition module, used to acquire multi-dimensional data of the target user when the target user is in a specific state, the multi-dimensional data including human body model data, environmental data, state intensity data and body surface temperature data;
[0044] A first processing module performs prediction processing on the multi-dimensional data based on a trained prediction model to obtain human body water loss data corresponding to the target user;
[0045] The second processing module is used to determine a water replenishment prompt strategy for the target user based on the human body water loss data, and send the water replenishment prompt strategy to the target user to prompt the user to replenish water according to the water replenishment prompt strategy.
[0046] In a third aspect, an embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the hydration reminder method provided in the embodiment of the present invention when executing the computer program.
[0047] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the water replenishment reminder method provided in the embodiment of the invention are implemented.
[0048] In an embodiment of the present invention, when a target user is in a specific state, multi-dimensional data of the target user is obtained, and the multi-dimensional data includes human body model data, environmental data, state intensity data, and body surface temperature data; based on a trained prediction model, the multi-dimensional data is predicted and processed to obtain human body water loss data corresponding to the target user; based on the human body water loss data, a hydration prompt strategy for the target user is determined, and the hydration prompt strategy is sent to the target user to prompt the user to replenish water according to the hydration prompt strategy. The present invention can use the multi-dimensional data of the user to make predictions when the user is in a specific state, predict the user's human body water loss data, and determine the hydration prompt strategy for the user based on the human body water loss data, so as to timely and accurately provide hydration prompts to the target user and improve the user's hydration experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0050] Figure 1 is a flow chart of a water replenishment prompting method provided by an embodiment of the present invention;
[0051] Figure 2 It is a structural schematic diagram of a water replenishment reminder device provided by an embodiment of the present invention;
[0052] Figure 3 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0054] like Figure 1 As shown, Figure 1 This is a method flow chart of a water replenishment reminder method provided by an embodiment of the present invention. The water replenishment reminder method comprises the steps of:
[0055] 101. When the target user is in a specific state, obtain multi-dimensional data of the target user.
[0056] In an embodiment of the present invention, the above-mentioned hydration reminder method can be applied to a smart wearable device, a mobile terminal device or a server. The above-mentioned target user is a user who needs a hydration reminder, specifically a user of a smart wearable device or a mobile terminal device, or a user who subscribes to a hydration reminder in the server.
[0057] The above-mentioned specific state can be a working state, a motion state, etc., or a user-defined state. The embodiment of the present invention takes the motion state as an example for explanation, and it can be understood that other types of specific states can also refer to the specific implementation of the motion state.
[0058] The above-mentioned multi-dimensional data can be collected through sensors in smart wearable devices or sensors in mobile terminal devices. For server subscription, the server can obtain the corresponding multi-dimensional data from the smart wearable device or mobile terminal device of the target user. Placing the calculation of multi-dimensional data in the server can reduce the hardware requirements of smart wearable devices and mobile terminal devices.
[0059] The multi-dimensional data may include human body model data, environmental data, state intensity data, and body surface temperature data. The human body model data may include the height, weight, and body shape of the target user. The body shape may be the size of each part of the body, such as chest circumference, waist circumference, hip circumference, etc., or may be a body type classification, such as thin, average, obese, etc. The environmental data may include wind speed, ambient temperature, ambient humidity, etc. The state intensity data may include respiratory rate, heart rate, etc. The body surface temperature data may include exhaled gas temperature, skin temperature, etc.
[0060] The above human body model data can be obtained by user upload. Environmental data can be obtained based on the built-in sensors of the smart wearable device, or through data sharing methods such as cloud weather forecasts. The above state intensity data can be collected and determined by the state sensor of the smart wearable device. The above state sensor can be a motion sensor, such as an inertial measurement unit, an acceleration sensor, an angular velocity sensor, etc. The above body surface temperature data can be collected and obtained by the built-in temperature sensor of the smart wearable device.
[0061] 102. Based on the trained prediction model, the multi-dimensional data is predicted and processed to obtain the human body water loss data corresponding to the target user.
[0062] In an embodiment of the present invention, the above-mentioned prediction model can be a machine learning model based on a convolutional neural network RNN, and can be a machine learning model based on a convolutional neural network such as LeNet, ResNet, VGGNet, yolo series, etc.
[0063] The above-mentioned prediction model can be a model trained based on multi-dimensional data. The output of the above-mentioned prediction model is human body water loss data. After the multi-dimensional data of the target user is input into the trained prediction model, the prediction model outputs the prediction result corresponding to the multi-dimensional data, and the prediction result is the human body water loss data corresponding to the target user.
[0064] The above-mentioned human body water loss data can be the amount of water lost through the human skin, and can also be understood as the amount of sweat. The above-mentioned human body water loss data can also be the sweat distribution of the human skin. For human skin, even under the same specific state, the corresponding sweat amount of each skin area is different. This is mainly due to the different distribution of human sweat glands. For example, the amount of sweating on the palms and the backs of the hands is different during exercise, and the amount of sweating on the forehead and shoulders is different. Therefore, the above-mentioned prediction model can output the sweat distribution of the human skin as the human body water loss data corresponding to the target user. The sweat distribution of the human skin can more accurately describe the sweating situation of the target user.
[0065] Sample multi-dimensional data can be collected, and the sample multi-dimensional data includes sample human body model data, sample environmental data, sample state intensity data, and sample body surface temperature data, and human body water loss data corresponding to the sample multi-dimensional data is collected as a label. A training data set is constructed using the collected sample multi-dimensional data and the corresponding labels. The training model is trained with the training data set, and the model is verified and optimized to obtain a trained prediction model. When the current specific values (human body model data, environmental data, state intensity data, and body surface temperature data) are input, the output value is the human body water loss data.
[0066] 103. Based on the human body water loss data, determine a water replenishment reminder strategy for the target user, and send the water replenishment reminder strategy to the target user to prompt the user to replenish water according to the water replenishment reminder strategy.
[0067] In the embodiment of the present invention, after the human body water loss data is obtained, a water replenishment prompt strategy for the target user can be determined according to the corresponding relationship between the human body water loss data and the water replenishment prompt strategy.
[0068] The above-mentioned water replenishment reminder strategy may include a water replenishment timing and a water replenishment amount. The above-mentioned water replenishment timing is used to remind the target user at which time to replenish water, and the above-mentioned water replenishment amount is used to remind the target user the amount of water to replenish each time.
[0069] After obtaining the hydration reminder strategy, the hydration reminder strategy can be sent to the target user through voice, information, etc., and the hydration reminder strategy can also be combined with the memo in the smart wearable device or mobile terminal device so that the hydration reminder strategy can be used to remind the target user through the memo.
[0070] In a possible embodiment, the above-mentioned water replenishment reminder strategy may also include navigation information. The above-mentioned human body water loss data is the amount of sweat. If the amount of sweat reaches a certain level, the target user is reminded to replenish electrolyte water to avoid dehydration. When the amount of sweat reaches a certain sweat threshold, the target user is recommended to drink electrolyte water. The address and distance of multiple nearby small stores, navigation and other information may be sent to the target user for the target user to choose and buy electrolyte water.
[0071] In an embodiment of the present invention, when a target user is in a specific state, multi-dimensional data of the target user is obtained, and the multi-dimensional data includes human body model data, environmental data, state intensity data, and body surface temperature data; based on a trained prediction model, the multi-dimensional data is predicted and processed to obtain human body water loss data corresponding to the target user; based on the human body water loss data, a hydration prompt strategy for the target user is determined, and the hydration prompt strategy is sent to the target user to prompt the user to replenish water according to the hydration prompt strategy. The present invention can use the multi-dimensional data of the user to make predictions when the user is in a specific state, predict the user's human body water loss data, and determine the hydration prompt strategy for the user based on the human body water loss data, so as to timely and accurately provide hydration prompts to the target user and improve the user's hydration experience.
[0072] It should be noted that the water replenishment reminder method provided in the embodiment of the present invention can be applied to devices such as smart cameras, smart phones, tablets, computers, servers, etc. that can perform the water replenishment reminder method.
[0073] It can be understood that in the specific implementation of this application, user data, environmental data, status data, vital sign data and other related data are involved. When the embodiments in this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of relevant data, as well as model training, deployment and use, must comply with relevant laws, regulations and standards of relevant countries and regions.
[0074] Optionally, the prediction model includes a human surface area feature extraction network, an environmental feature extraction network, a state intensity feature extraction network, a sweat evaporation feature extraction network, a feature fusion network and an output network. In the step of predicting and processing the multi-dimensional data based on the trained prediction model to obtain the human body water loss data corresponding to the target user, the human body model data can be feature extracted by the human surface area feature extraction network to obtain the human body surface features corresponding to the target user; the environmental data can be feature extracted by the environmental feature extraction network to obtain the environmental features corresponding to the target user; the state intensity data can be feature extracted by the state intensity feature extraction network to obtain the state intensity features corresponding to the target user; the body surface temperature data can be feature extracted by the sweat evaporation feature extraction network to obtain the sweat evaporation features corresponding to the target user; the human body surface features, environmental features, state intensity features and sweat evaporation features can be fused by the fusion network to obtain the fusion features of the target user; the fusion features can be linearly regressed by the output network to output the human body water loss data corresponding to the target user.
[0075] In an embodiment of the present invention, the human body surface area feature extraction network, the environmental feature extraction network, the state intensity feature extraction network, and the sweat evaporation amount feature extraction network are parallel feature extraction networks, and the output ends of the human body surface area feature extraction network, the environmental feature extraction network, the state intensity feature extraction network, and the sweat evaporation amount feature extraction network are all connected to the input end of the feature fusion network, and the output end of the feature fusion network is connected to the input end of the output network, and the output of the output network is human body water loss data.
[0076] The input of the human body surface area feature extraction network is human body model data, and the human body surface area feature extraction network is used to extract human body surface area features. The human body surface features can be understood as an abstract data representation, which does not directly represent the size and / or shape of the human body surface area, but implicitly represents the size and / or shape of the human body surface area.
[0077] In a possible embodiment, for a user, the human body surface area feature is fixed. After the human body surface area feature of the target user is extracted based on the human body model data of the target user, the extracted human body surface area feature can be stored and updated regularly or when at least one of the height, weight and body shape of the target user is detected to have changed. Therefore, the human body surface area feature extraction network can adopt a normally closed strategy. When reusing the stored human body surface area features, the human body surface area feature extraction network is in a closed state, and no human body model data needs to be input. When an update is required, the human body surface area feature extraction network is opened to re-extract the human body surface area features.
[0078] The input of the above-mentioned environmental feature extraction network is environmental data. The above-mentioned environmental feature extraction network is used to extract the environmental features of the target user's current environment. The above-mentioned environmental features can be understood as an abstract data representation used to implicitly represent environmental information.
[0079] The input of the state strength feature extraction network is the state strength data. The state strength feature extraction network is used to extract the state strength of the target user in a specific state. The stronger the state strength, the greater the human body consumption, and the weaker the state strength, the smaller the human body consumption. The state strength feature can be understood as an abstract data representation, which is used to implicitly represent the state strength.
[0080] The input of the sweat evaporation feature extraction network is the body surface temperature data. The sweat evaporation feature extraction network is used to extract the sweat evaporation feature of the target user. The sweat evaporation feature can be understood as an abstract data representation, which is used to implicitly represent the user's sweat evaporation. It should be noted that the sweat evaporation can be calculated by the skin temperature difference. The difference between the skin temperature and the ambient temperature (i.e., the skin temperature difference) determines the amount and speed of sweating. The greater the skin temperature difference, the more sweat the human body will secrete to accelerate evaporation and heat dissipation in order to maintain a stable body temperature.
[0081] The feature fusion network is used to fuse the human body surface features, environmental features, state intensity features, and sweat evaporation features, so that the fused features can include feature information of multiple dimensions, thereby improving the prediction accuracy. In the feature fusion network, the human body surface features, environmental features, state intensity features, and sweat evaporation features can be regularized (linearly changed) so that the human body surface features, environmental features, state intensity features, and sweat evaporation features have the same dimension, and feature fusion is performed under the same dimension. The feature fusion method can be channel superposition or feature splicing.
[0082] The output network is used to perform linear regression processing on the fusion features, thereby outputting corresponding human body water loss data. The output network may include a fully connected layer and a classification layer, wherein the fully connected layer includes a plurality of neurons, and the plurality of neurons are a fully connected structure, and the classification layer includes classification parameters, and the output of the fully connected structure is classified by the classification parameters, thereby obtaining human body water loss data.
[0083] By extracting human body surface features, environmental features, state intensity features, and sweat evaporation features, human body water loss data can be predicted more comprehensively and accurately, thereby improving the accuracy of human body water loss data.
[0084] Optionally, before performing feature extraction processing on the human body model data through the human body surface area feature extraction network to obtain the human body surface features corresponding to the target user, first training sample data and a three-dimensional human body generation model can also be obtained, the first training sample data including first sample human body model data and a three-dimensional human body label, one sample human body model data corresponds to one three-dimensional human body label; the three-dimensional human body generation model is trained through the first training sample data to obtain a trained three-dimensional human body generation model; based on the trained three-dimensional human body generation model, a human body surface area feature extraction network is generated.
[0085] In an embodiment of the present invention, the human body surface area feature extraction network can be constructed based on a trained three-dimensional human body generation model. Specifically, the human body model data can be processed by the trained three-dimensional human body generation model to obtain a three-dimensional human body model, and the human body surface area feature can be determined based on the surface of the three-dimensional human body model, thereby constructing a mapping path between the human body model data and the human body surface area feature, and generating a human body surface area feature extraction network based on the mapping path.
[0086] The data structure of the first sample human body model data is the same as the data structure of the human body model data, that is, if the human body model data includes height, weight, body shape, etc., then the first sample human body model data also includes height, weight, body shape, etc. The three-dimensional human body label can be understood as a verified three-dimensional human body data, which is used to guide the first sample human body model data to generate a three-dimensional human body model that is the same or similar to the three-dimensional human body label.
[0087] The above-mentioned training method can adopt a supervised training method, and the first sample human body model data is generated and processed through the three-dimensional human body generation model to obtain a generated three-dimensional human body model. The error value between the generated three-dimensional human body model and the three-dimensional human body label is calculated through a preset error function. The model parameters in the three-dimensional human body generation model are adjusted with the goal of minimizing the error value. After iterating a preset number of times, the iteration is stopped, and a trained three-dimensional human body generation model can be obtained.
[0088] Optionally, in the step of generating a human body surface area feature extraction network based on a trained three-dimensional human body generation model, a first three-dimensional human body model corresponding to the first sample human body model data can be generated by the trained three-dimensional human body generation model, and each first sample human body model data corresponds to a first three-dimensional human body model; surface segmentation processing is performed on the first three-dimensional human body model to obtain a first human body surface area feature corresponding to the first three-dimensional human body model; a mapping network between the first sample human body model data and the first human body surface area feature is constructed; and a human body surface area feature extraction network is generated based on the mapping network.
[0089] In the embodiment of the present invention, after obtaining the trained 3D human body generation model, the first sample human body model data is generated and processed by the trained 3D human body generation model to obtain a first 3D human body model corresponding to the first sample human body model data.
[0090] The surface of the first three-dimensional human body model can be decomposed into smaller parts (such as triangles or quadrilaterals), and the sum of the areas of these parts is calculated to obtain the total surface area of the first three-dimensional human body model. The above surface area feature can be a sequence of total surface area or partial surface areas. In the sequence of partial surface areas, each decomposed part corresponds to a number, and the surface areas of the parts are sorted according to the numbers to obtain a sequence of partial surface areas.
[0091] After obtaining the first human body surface area feature corresponding to the first three-dimensional human body model, a mapping network between the first sample human body model data and the first human body surface area feature can be constructed, and then the mapping network can be determined as a human body surface area feature extraction network.
[0092] By generating a corresponding first three-dimensional human body model through the first sample human body model data and utilizing the mapping relationship between the first three-dimensional human body model and the human body surface area features, the human body surface area feature extraction network can make full use of the modeling data of the three-dimensional human body model, thereby extracting more accurate human body surface area features.
[0093] Optionally, in the step of generating a human body surface area feature extraction network based on a trained three-dimensional human body generation model, second training sample data and a human body surface segmentation model can be obtained, the second training sample data including second sample human body model data, a second three-dimensional human body model, a human body surface sweat gland distribution map, and a human body surface sweating area label, the second three-dimensional human body model is generated by processing the trained three-dimensional human body generation model with the second sample human body model data; the human body surface segmentation model is trained based on the second training sample data to obtain a trained human body surface segmentation model; based on the trained human body surface segmentation model, a human body surface area feature extraction network is constructed.
[0094] In an embodiment of the present invention, the data structure of the second sample human body model data is the same as the data structure of the human body model data, that is, the human body model data includes height, weight, body shape, etc., and the second sample human body model data also includes height, weight, body shape, etc. The human body surface sweating area label can be understood as a verified human body surface sweating area distribution data, which is used to guide the human body surface segmentation model to segment the second three-dimensional human body model according to the human body surface sweating area, and obtain a human body surface area feature whose sweating area distribution is close to the real human body sweating area distribution.
[0095] The second sample human body model data is input into the trained three-dimensional human body generation model to obtain a second three-dimensional human body model. The human body surface sweat gland distribution map can be pre-prepared standard data and can be obtained on various medical platforms. The second three-dimensional human body model and the human body surface sweat gland distribution map are auxiliary data. After the training is completed, the second three-dimensional human body model and the human body surface sweat gland distribution map are no longer needed as inputs to the human body surface area feature extraction network.
[0096] The second training sample data is input into the human body surface segmentation model, and the output is human body surface sweat area distribution data with human body surface sweat area labels, and the human body surface sweat area distribution data can be heat map data. The human body surface sweat area distribution data is used as the human body surface area feature. Then, a mapping relationship between the second sample human body model data and the human body surface area feature can be established, and the expression of the mapping relationship is the trained human body surface segmentation model, and then the trained human body surface segmentation model can be directly used as a human body surface area feature extraction network, and only the human body model data needs to be input to output the human body surface area feature.
[0097] Optionally, the above-mentioned environmental data includes wind data, temperature data and humidity data. In the step of performing feature extraction processing on the environmental data through the environmental feature extraction network to obtain the environmental features corresponding to the target user, the wind data, temperature data and humidity data can be subjected to feature extraction processing through the environmental feature extraction network to obtain the environmental features corresponding to the target user. In the step of performing feature extraction processing on the state intensity data through the state intensity feature extraction network to obtain the state intensity features corresponding to the target user, the state intensity value corresponding to the target user can be determined based on the state intensity data and the reference state data; the state intensity value can be subjected to feature extraction processing through the state intensity feature extraction network to obtain the state intensity features corresponding to the target user. The above-mentioned body surface temperature data includes skin temperature data and environmental temperature data. In the step of performing feature extraction processing on the body surface temperature data through the sweat evaporation feature extraction network to obtain the sweat evaporation features corresponding to the target user, the skin temperature difference of the target user can be determined based on the difference between the skin temperature data and the environmental temperature data; the skin temperature difference can be subjected to feature extraction processing through the sweat evaporation feature extraction network to obtain the sweat evaporation features corresponding to the target user.
[0098] In an embodiment of the present invention, wind data, temperature data, and humidity data can be obtained from sensors built into smart wearable devices, or can be obtained through data sharing methods such as cloud weather forecasts. The above-mentioned environmental feature extraction network performs feature extraction processing on wind data, temperature data, and humidity data, which can be understood as abstracting wind data, temperature data, and humidity data, so that specific environmental data can be abstracted, such as being abstracted into a feature matrix or feature vector, which can facilitate machine calculation.
[0099] The larger the above state intensity value is, the greater the consumption of the human body in the corresponding specific state is, and the smaller the state intensity is, the smaller the consumption of the human body in the corresponding specific state is.
[0100] Taking the exercise state as an example, the exercise intensity value can be calculated based on the heart rate obtained by the smart wearable device. Heart rate is one of the important indicators for measuring exercise intensity. The greater the exercise intensity, the higher the heart rate is usually. Smart wearable devices continuously track the user's heart rate data through a built-in heart rate monitor. The above heart rate data may include exercise heart rate, resting heart rate, and maximum heart rate. By calculation formula: exercise intensity value (%) = (exercise heart rate - resting heart rate) ÷ (maximum heart rate - resting heart rate) × 100%. The above formula takes into account the resting heart rate, and the exercise intensity value can more accurately reflect the individual's exercise intensity.
[0101] The above-mentioned state strength feature extraction network extracts features from the state strength value, which can be understood as abstracting the state strength value so that the specific state strength value can be represented abstractly, such as abstracted as a feature matrix or feature vector, which is convenient for machine calculation.
[0102] Skin temperature data and ambient temperature data can be collected and obtained through the temperature sensor built into the smart wearable device. Based on the difference between the skin temperature data and the ambient temperature data, the skin temperature difference of the target user is determined. If the difference is negative, it means that the skin temperature is lower than the ambient temperature, and if the difference is positive, it means that the skin temperature is higher than the ambient temperature.
[0103] The sweat evaporation feature extraction network performs feature extraction processing on the skin temperature difference, which can be understood as abstracting the skin temperature difference so that the specific skin temperature difference can be represented abstractly, such as abstracted as a feature matrix or feature vector, which is convenient for machine calculation.
[0104] Optionally, in the step of determining a hydration reminder strategy for a target user based on human body water loss data, the target user's body water loss amount can be determined based on the human body water loss data; the target user's total water replenishment amount can be determined based on the human body water loss amount; the target user's water replenishment amount at each hydration is determined based on the target user's hydration habits and the total water replenishment amount, wherein the hydration habits include the frequency of hydration and the type of hydration; and a hydration reminder strategy for the target user is determined based on the target user's water replenishment amount at each hydration.
[0105] In an embodiment of the present invention, the human body water loss data may be a sweat amount, and the target user's water replenishment amount is calculated based on the target user's sweat amount. The water replenishment amount may be n times the sweat amount, where n is a value greater than 0, and n may be set based on experience, such as 1.5, that is, the water replenishment amount is 1.5 times the sweat amount.
[0106] The above-mentioned hydration type can be understood as the type of water that the user drinks, which can be different types such as beverages and purified water. Different hydration types can correspond to different type coefficients. After the hydration amount is calculated, the hydration amount is corrected according to the hydration type. For example, if the type coefficient of purified water is 1, the hydration amount remains unchanged. If the type coefficient of carbonated beverages is 1.2, the hydration amount is 1.2 times the original.
[0107] According to the user's habits, the amount of water they drink each time can be calculated against the calculated amount of water they drink, and the number of reminders and the interval between reminders can be obtained. After the interval time is reached, the user can be reminded subsequently.
[0108] In a possible embodiment, in the water replenishment reminder strategy, after a specific state ends, the rule of "drinking small amounts frequently" can be prompted to remind the user to drink water; the amount of water consumed can be calculated, and the amount of water consumed can be evenly divided into multiple recommended amounts of water consumed according to the principle of "drinking small amounts frequently", and the target user can be reminded at regular intervals so that the user can replenish water according to the recommended amount of water consumed.
[0109] In a possible embodiment, if the amount of sweat reaches a certain level, a reminder to drink electrolyte water is given to avoid dehydration. When the amount of sweat reaches a certain sweat threshold, the target user is recommended to drink electrolyte water, and the addresses, distances, navigation information, etc. of multiple nearby small stores can be sent to the target user for the target user to choose and buy electrolyte water, so that the target user can replenish electrolytes.
[0110] like Figure 2 As shown, an embodiment of the present invention provides a water replenishment reminder device, which includes:
[0111] An acquisition module 201 is used to acquire multi-dimensional data of the target user when the target user is in a specific state, wherein the multi-dimensional data includes human body model data, environmental data, state intensity data, and body surface temperature data;
[0112] The first processing module 202 performs prediction processing on the multi-dimensional data based on the trained prediction model to obtain the human body water loss data corresponding to the target user;
[0113] The second processing module 203 is used to determine a water replenishment prompt strategy for the target user based on the human body water loss data, and send the water replenishment prompt strategy to the target user to prompt the user to replenish water according to the water replenishment prompt strategy.
[0114] Optionally, the prediction model includes a human surface area feature extraction network, an environmental feature extraction network, a state intensity feature extraction network, a sweat evaporation feature extraction network, a feature fusion network and an output network, and the first processing module 202 is also used to perform feature extraction processing on the human model data through the human surface area feature extraction network to obtain the human surface features corresponding to the target user; perform feature extraction processing on the environmental data through the environmental feature extraction network to obtain the environmental features corresponding to the target user; perform feature extraction processing on the state intensity data through the state intensity feature extraction network to obtain the state intensity features corresponding to the target user; perform feature extraction processing on the body surface temperature data through the sweat evaporation feature extraction network to obtain the sweat evaporation features corresponding to the target user; perform fusion processing on the human surface features, the environmental features, the state intensity features and the sweat evaporation features through the fusion network to obtain the fusion features of the target user; perform linear regression processing on the fusion features through the output network to output the human body water loss data corresponding to the target user.
[0115] Optionally, the device further comprises:
[0116] A data preparation module, used to obtain first training sample data and generate a three-dimensional human body model, wherein the first training sample data includes first sample human body model data and a three-dimensional human body label, and one sample human body model data corresponds to one three-dimensional human body label;
[0117] A training module, used for training the three-dimensional human body generation model using the first training sample data to obtain a trained three-dimensional human body generation model;
[0118] A generation module is used to generate the human body surface area feature extraction network based on the trained three-dimensional human body generation model.
[0119] Optionally, the generation module is also used to generate a first three-dimensional human body model corresponding to the first sample human body model data through the trained three-dimensional human body generation model, and each first sample human body model data corresponds to one first three-dimensional human body model; perform surface segmentation processing on the first three-dimensional human body model to obtain a first human body surface area feature corresponding to the first three-dimensional human body model; construct a mapping network between the first sample human body model data and the first human body surface area feature; and generate the human body surface area feature extraction network based on the mapping network.
[0120] Optionally, the generation module is also used to obtain second training sample data and a human body surface segmentation model, the second training sample data including second sample human body model data, a second three-dimensional human body model, a human body surface sweat gland distribution map and a human body surface sweating area label, the second three-dimensional human body model is generated by processing the trained three-dimensional human body generation model on the second sample human body model data; the human body surface segmentation model is trained based on the second training sample data to obtain a trained human body surface segmentation model; based on the trained human body surface segmentation model, the human body surface area feature extraction network is constructed.
[0121] Optionally, the environmental data includes wind data, temperature data and humidity data, and the first processing module 202 is further used to perform feature extraction processing on the wind data, the temperature data and the humidity data through the environmental feature extraction network to obtain the environmental features corresponding to the target user;
[0122] The first processing module 202 is further used to determine the state strength value corresponding to the target user based on the state strength data and the reference state data; perform feature extraction processing on the state strength value through the state strength feature extraction network to obtain the state strength feature corresponding to the target user;
[0123] The body surface temperature data includes skin temperature data and ambient temperature data. The first processing module 202 is also used to determine the skin temperature difference of the target user based on the difference between the skin temperature data and the ambient temperature data; perform feature extraction processing on the skin temperature difference through the sweat evaporation feature extraction network to obtain the sweat evaporation feature corresponding to the target user.
[0124] Optionally, the second processing module 203 is also used to determine the amount of human water loss of the target user based on the human water loss data; determine the total water replenishment amount of the target user based on the human water loss amount; determine the amount of water replenishment of the target user at each water replenishment based on the target user's water replenishment habits and the total water replenishment amount, the water replenishment habits including the frequency of water replenishment and the type of water replenishment; and determine a water replenishment reminder strategy for the target user based on the water replenishment amount of the target user at each water replenishment.
[0125] It should be noted that the hydration reminder device provided in the embodiment of the present invention can be applied to smart wearable devices, smart phones, tablets, computers, servers and other devices that can perform hydration reminder methods.
[0126] The water replenishment reminder device provided in the embodiment of the present invention can realize each process realized by the water replenishment reminder method in the above method embodiment, and can achieve the same beneficial effect. To avoid repetition, it will not be described here.
[0127] See also Figure 3 , Figure 3 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 3 As shown, it includes: a memory 302, a processor 301, and a computer program of a water replenishment prompt method stored in the memory 302 and executable on the processor 301, wherein:
[0128] The processor 301 is used to call the computer program stored in the memory 302 and execute the following steps:
[0129] When the target user is in a specific state, multi-dimensional data of the target user is obtained, wherein the multi-dimensional data includes human body model data, environmental data, state intensity data, and body surface temperature data;
[0130] Based on the trained prediction model, the multi-dimensional data is predicted and processed to obtain the human body water loss data corresponding to the target user;
[0131] Based on the human body water loss data, a water replenishment prompt strategy for the target user is determined, and the water replenishment prompt strategy is sent to the target user to prompt the user to replenish water according to the water replenishment prompt strategy.
[0132] Optionally, the prediction model includes a human surface area feature extraction network, an environmental feature extraction network, a state intensity feature extraction network, a sweat evaporation feature extraction network, a feature fusion network, and an output network. The processor 301 performs the step of performing prediction processing on the multi-dimensional data based on the trained prediction model to obtain the human body water loss data corresponding to the target user, including:
[0133] Performing feature extraction processing on the human body model data through the human body surface area feature extraction network to obtain human body surface features corresponding to the target user;
[0134] Performing feature extraction processing on the environmental data through the environmental feature extraction network to obtain environmental features corresponding to the target user;
[0135] Performing feature extraction processing on the state strength data through the state strength feature extraction network to obtain the state strength feature corresponding to the target user;
[0136] Performing feature extraction processing on the body surface temperature data through the sweat evaporation feature extraction network to obtain sweat evaporation features corresponding to the target user;
[0137] The human body surface features, the environmental features, the state intensity features, and the sweat evaporation amount features are fused through the fusion network to obtain fusion features of the target user;
[0138] The fused features are subjected to linear regression processing through the output network to output the human body water loss data corresponding to the target user.
[0139] Optionally, before the step of performing feature extraction processing on the human body model data through the human body surface area feature extraction network to obtain human body surface features corresponding to the target user, the method executed by the processor 301 further includes:
[0140] Acquire first training sample data and a three-dimensional human body generation model, wherein the first training sample data includes first sample human body model data and a three-dimensional human body label, and one sample human body model data corresponds to one three-dimensional human body label;
[0141] Training the three-dimensional human body generation model using the first training sample data to obtain a trained three-dimensional human body generation model;
[0142] Based on the trained three-dimensional human body generation model, the human body surface area feature extraction network is generated.
[0143] Optionally, the step of generating the human body surface area feature extraction network based on the trained three-dimensional human body generation model performed by the processor 301 includes:
[0144] Generate a first three-dimensional human body model corresponding to the first sample human body model data by using the trained three-dimensional human body generation model, wherein each first sample human body model data corresponds to one first three-dimensional human body model;
[0145] Performing surface segmentation processing on the first three-dimensional human body model to obtain a first human body surface area feature corresponding to the first three-dimensional human body model;
[0146] Constructing a mapping network between the first sample human body model data and the first human body surface area feature;
[0147] Based on the mapping network, the human body surface area feature extraction network is generated.
[0148] Optionally, the step of generating the human body surface area feature extraction network based on the trained three-dimensional human body generation model performed by the processor 301 includes:
[0149] Acquire second training sample data and a human body surface segmentation model, wherein the second training sample data includes second sample human body model data, a second three-dimensional human body model, a human body surface sweat gland distribution map, and a human body surface sweating area label, and the second three-dimensional human body model is generated by the trained three-dimensional human body generation model and processed by the second sample human body model data;
[0150] Training the human body surface segmentation model based on the second training sample data to obtain a trained human body surface segmentation model;
[0151] Based on the trained human body surface segmentation model, the human body surface area feature extraction network is constructed.
[0152] Optionally, the environmental data includes wind data, temperature data, and humidity data, and the step of performing feature extraction processing on the environmental data through the environmental feature extraction network to obtain the environmental features corresponding to the target user, performed by the processor 301, includes:
[0153] Performing feature extraction processing on the wind data, the temperature data, and the humidity data through the environmental feature extraction network to obtain environmental features corresponding to the target user;
[0154] The step of performing feature extraction processing on the state strength data through the state strength feature extraction network to obtain the state strength feature corresponding to the target user, performed by the processor 301, includes:
[0155] Determine a state intensity value corresponding to the target user based on the state intensity data and the reference state data;
[0156] Performing feature extraction processing on the state strength value through the state strength feature extraction network to obtain the state strength feature corresponding to the target user;
[0157] The body surface temperature data includes skin temperature data and ambient temperature data. The processor 301 performs feature extraction processing on the body surface temperature data through the sweat evaporation feature extraction network to obtain the sweat evaporation feature corresponding to the target user, including:
[0158] Determining a skin temperature difference of the target user based on a difference between the skin temperature data and the ambient temperature data;
[0159] The sweat evaporation feature extraction network performs feature extraction processing on the skin temperature difference to obtain the sweat evaporation feature corresponding to the target user.
[0160] Optionally, the step of determining a water replenishment prompt strategy for the target user based on the human body water loss data performed by the processor 301 includes:
[0161] Determining the body water loss amount of the target user based on the body water loss data;
[0162] Determining a total water replenishment amount for the target user based on the water loss of the human body;
[0163] Determining the water replenishment amount of the target user at each water replenishment based on the water replenishment habit of the target user and the total water replenishment amount, wherein the water replenishment habit includes the water replenishment frequency and the water replenishment type;
[0164] Based on the amount of water replenished by the target user during each water replenishment, a water replenishment prompt strategy for the target user is determined.
[0165] It should be noted that the electronic device provided in the embodiment of the present invention can be applied to smart wearable devices, smart phones, tablets, computers, servers and other devices that can perform the water replenishment reminder method.
[0166] The electronic device provided in the embodiment of the present invention can implement each process implemented by the water replenishment prompt method in the above method embodiment, and can achieve the same beneficial effect. To avoid repetition, it will not be described here.
[0167] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the water replenishment reminder method provided in the embodiment of the present invention are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
[0168] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The computer-readable storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).
[0169] The above disclosure is only the preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.
Claims
1. A water replenishment reminder method, characterized in that: The method comprises the following steps: When the target user is in a specific state, multi-dimensional data of the target user is obtained, wherein the multi-dimensional data includes human body model data, environmental data, state intensity data, and body surface temperature data; Based on the trained prediction model, the multi-dimensional data is predicted and processed to obtain the human body water loss data corresponding to the target user; Based on the human body water loss data, a water replenishment prompt strategy for the target user is determined, and the water replenishment prompt strategy is sent to the target user to prompt the target user to replenish water according to the water replenishment prompt strategy.
2. The water replenishment reminder method according to claim 1, characterized in that: The prediction model includes a human surface area feature extraction network, an environmental feature extraction network, a state intensity feature extraction network, a sweat evaporation feature extraction network, a feature fusion network, and an output network. The step of performing prediction processing on the multi-dimensional data based on the trained prediction model to obtain the human body water loss data corresponding to the target user includes: Performing feature extraction processing on the human body model data through the human body surface area feature extraction network to obtain human body surface features corresponding to the target user; Performing feature extraction processing on the environmental data through the environmental feature extraction network to obtain environmental features corresponding to the target user; Performing feature extraction processing on the state strength data through the state strength feature extraction network to obtain the state strength feature corresponding to the target user; Performing feature extraction processing on the body surface temperature data through the sweat evaporation feature extraction network to obtain sweat evaporation features corresponding to the target user; The human body surface features, the environmental features, the state intensity features, and the sweat evaporation amount features are fused through the fusion network to obtain fusion features of the target user; The fused features are subjected to linear regression processing through the output network to output the human body water loss data corresponding to the target user.
3. The water replenishment reminder method according to claim 2, characterized in that: Before the step of performing feature extraction processing on the human body model data through the human body surface area feature extraction network to obtain the human body surface features corresponding to the target user, the method further includes: Acquire first training sample data and a three-dimensional human body generation model, wherein the first training sample data includes first sample human body model data and a three-dimensional human body label, and one sample human body model data corresponds to one three-dimensional human body label; Training the three-dimensional human body generation model using the first training sample data to obtain a trained three-dimensional human body generation model; Based on the trained three-dimensional human body generation model, the human body surface area feature extraction network is generated.
4. The water replenishment reminder method according to claim 3, characterized in that: The step of generating the human body surface area feature extraction network based on the trained three-dimensional human body generation model comprises: Generate a first three-dimensional human body model corresponding to the first sample human body model data by using the trained three-dimensional human body generation model, wherein each first sample human body model data corresponds to one first three-dimensional human body model; Performing surface segmentation processing on the first three-dimensional human body model to obtain a first human body surface area feature corresponding to the first three-dimensional human body model; Constructing a mapping network between the first sample human body model data and the first human body surface area feature; Based on the mapping network, the human body surface area feature extraction network is generated.
5. The water replenishment reminder method according to claim 3, characterized in that: The step of generating the human body surface area feature extraction network based on the trained three-dimensional human body generation model comprises: Acquire second training sample data and a human body surface segmentation model, wherein the second training sample data includes second sample human body model data, a second three-dimensional human body model, a human body surface sweat gland distribution map, and a human body surface sweating area label, and the second three-dimensional human body model is generated by the trained three-dimensional human body generation model and processed by the second sample human body model data; Training the human body surface segmentation model based on the second training sample data to obtain a trained human body surface segmentation model; Based on the trained human body surface segmentation model, the human body surface area feature extraction network is constructed.
6. The water replenishment reminder method according to claim 2, characterized in that: The environmental data includes wind data, temperature data and humidity data, and the step of performing feature extraction processing on the environmental data through the environmental feature extraction network to obtain the environmental features corresponding to the target user includes: Performing feature extraction processing on the wind data, the temperature data, and the humidity data through the environmental feature extraction network to obtain environmental features corresponding to the target user; The step of performing feature extraction processing on the state strength data through the state strength feature extraction network to obtain the state strength feature corresponding to the target user includes: Determine a state intensity value corresponding to the target user based on the state intensity data and the reference state data; Performing feature extraction processing on the state strength value through the state strength feature extraction network to obtain the state strength feature corresponding to the target user; The body surface temperature data includes skin temperature data and ambient temperature data, and the step of performing feature extraction processing on the body surface temperature data through the sweat evaporation feature extraction network to obtain the sweat evaporation feature corresponding to the target user includes: Determining a skin temperature difference of the target user based on a difference between the skin temperature data and the ambient temperature data; The sweat evaporation feature extraction network performs feature extraction processing on the skin temperature difference to obtain the sweat evaporation feature corresponding to the target user.
7. The water replenishment reminder method according to any one of claims 1 to 6, characterized in that: The step of determining a water replenishment prompt strategy for the target user based on the human body water loss data comprises: Determining the body water loss amount of the target user based on the body water loss data; Determining a total water replenishment amount for the target user based on the water loss of the human body; Determining the water replenishment amount of the target user at each water replenishment based on the water replenishment habit of the target user and the total water replenishment amount, wherein the water replenishment habit includes the water replenishment frequency and the water replenishment type; Based on the amount of water replenished by the target user during each water replenishment, a water replenishment prompt strategy for the target user is determined.
8. A water replenishment reminder device, characterized in that: The water replenishment prompting device comprises: An acquisition module, used to acquire multi-dimensional data of the target user when the target user is in a specific state, the multi-dimensional data including human body model data, environmental data, state intensity data and body surface temperature data; A first processing module performs prediction processing on the multi-dimensional data based on a trained prediction model to obtain human body water loss data corresponding to the target user; The second processing module is used to determine a water replenishment prompt strategy for the target user based on the human body water loss data, and send the water replenishment prompt strategy to the target user to prompt the user to replenish water according to the water replenishment prompt strategy.
9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the hydration reminder method as described in any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the water replenishment reminder method according to any one of claims 1 to 7 are implemented.
