Wounded person positioning method and system based on neural network prediction

Abstract

The invention discloses a wounded person positioning method and system based on neural network prediction. The method comprises the following steps: acquiring sensing data of a plurality of wounded persons in an accident area; inputting the sensing data into a first prediction network to obtain the injury condition of each wounded person; the first prediction network is arranged in edge equipment of the accident area; screening out a plurality of to-be-positioned wounded persons of which the injury conditions are greater than a preset condition threshold value from all the wounded persons; and inputting the sensing data of the to-be-positioned wounded person into a second prediction network arranged in cloud equipment to obtain position information and injury condition information corresponding to the to-be-positioned wounded person. Therefore, the method can achieve the efficient and accurate screening and positioning of the wounded, guarantees the priority distribution of resources to the critical wounded in the subsequent rescue, optimizes the rescue scheduling efficiency, and improves the overall rescue effect.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for locating a wounded person based on neural network prediction. Background Art

[0002] In existing accident emergency rescue technologies, the assessment and location of the injured usually rely on manual judgment or traditional monitoring equipment. In traditional methods, rescuers need to check the status of the injured one by one, or rely on fixed monitoring equipment to obtain injury information. However, these methods are difficult to quickly cover the entire accident area in the early stages after the accident, and are prone to omissions or misjudgments. In addition, due to the lack of an efficient data prediction mechanism, the existing injured positioning technology is difficult to quickly screen the location of the injured who need priority treatment, which in turn affects the timeliness and accuracy of the overall rescue. It can be seen that the existing technology has defects that need to be solved urgently. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a method and system for locating the wounded based on neural network prediction, which can realize efficient and accurate screening and locating of the wounded, ensure that resources are allocated preferentially to critically injured persons in subsequent rescue, optimize rescue scheduling efficiency, and improve the overall rescue effect.

[0004] In order to solve the above technical problems, the first aspect of the present invention discloses a method for locating a wounded person based on neural network prediction, the method comprising: Acquire sensor data of multiple casualties in the accident area; Inputting the sensor data into a first prediction network to obtain the injury status of each of the injured persons; the first prediction network is arranged in an edge device in the accident area; Screening out a plurality of injured persons to be located whose injury conditions are greater than a preset threshold value from all the injured persons; The sensor data of the injured person to be located is input into a second prediction network arranged in a cloud device to obtain the position information and injury information corresponding to the injured person to be located.

[0005] As an optional embodiment, in the first aspect of the present invention, the sensor data is obtained by a drone device equipped with a sensor module installed in the accident area and / or a sensor module installed on the injured person.

[0006] As an optional embodiment, in the first aspect of the present invention, the sensor data includes at least one of physiological data, motion data, environmental sensor data, image data and sound data; the physiological data includes at least one of heart rate, blood oxygen saturation, blood pressure, body temperature, respiratory rate, electrocardiogram data, blood sugar level and bleeding detection data; the motion data includes at least one of acceleration, angular velocity, displacement information and limb activity; the environmental sensor data includes at least one of temperature, humidity, air quality, dust concentration and light intensity.

[0007] As an optional embodiment, in the first aspect of the present invention, the first prediction network is trained by a training data set including multiple training casualty sensor data and corresponding injury annotations; the second prediction network is trained by a training data set including multiple training casualty sensor data and corresponding position annotations and injury annotations; the model parameter scale of the second prediction network is greater than the model parameter scale of the first prediction network.

[0008] As an optional implementation, in the first aspect of the present invention, the first prediction network and the second prediction network are trained by the following steps: Acquire a common training data set including sensor data of multiple training injured persons and corresponding injury annotations; Based on the common training data set, the preset prediction network basic model is trained until convergence to obtain a trained common prediction network model; Based on a preset distillation target parameter, distilling the common prediction network model to obtain the first prediction network; Acquire an advanced training data set including sensor data of multiple training injured persons and corresponding position annotations and injury condition annotations; The common prediction network model is trained according to the advanced training data set until convergence to obtain the second prediction network.

[0009] As an optional implementation, in the first aspect of the present invention, the distilling the common prediction network model based on a preset distillation target parameter to obtain the first prediction network includes: Obtain device parameters of edge devices in the accident area; Determine the model carrying capacity parameters corresponding to the equipment parameters based on a preset parameter correspondence analysis algorithm; Inputting the model carrying capacity parameter into the model scale prediction network corresponding to the network architecture corresponding to the prediction network basic model to obtain the prediction model scale; the network architecture includes at least one of CNN architecture, RNN architecture, LSTM architecture and random forest architecture; Determining distillation target parameters according to the prediction model scale and the current model scale of the common prediction network model, and the correspondence between the preset scale and the distillation parameter; the distillation target parameters include a distillation target function value, a distillation constraint condition function, and a distillation operation duration; The common prediction network model is distilled according to the distillation target parameter to obtain the first prediction network.

[0010] As an optional implementation, in the first aspect of the present invention, the inputting the sensor data of the injured person to be located into a second prediction network set in a cloud device to obtain the location information and injury information corresponding to the injured person to be located includes: For each of the injured persons to be located, the sensor data of the injured person to be located is transmitted to the cloud device, and the acquisition location and acquisition time of the sensor data and the receiving time of the cloud device are recorded; Inputting the sensor data into a second prediction network arranged in the cloud device to obtain a first predicted position and a predicted injury degree corresponding to the injured person to be located; Inputting the acquisition time and the reception time into a trained position prediction network to obtain a second predicted position; the position prediction network is trained by a training data set including a plurality of training transmission time information and corresponding position annotations; The position information corresponding to the injured person to be located is determined according to the acquired position, the first predicted position and the second predicted position.

[0011] As an optional implementation, in the first aspect of the present invention, determining the position information corresponding to the injured person to be located according to the acquired position, the first predicted position and the second predicted position includes: Calculating a position correction weight proportional to the predicted injury severity; The product of the position correction weight and the acquired position and the average value of the product, the first predicted position and the second predicted position are calculated to obtain the position information corresponding to the injured person to be located.

[0012] A second aspect of an embodiment of the present invention discloses a system for locating a wounded person based on neural network prediction, the system comprising: An acquisition module, used to acquire sensor data of multiple injured persons in the accident area; A first prediction module, used for inputting the sensor data into a first prediction network to obtain the injury condition of each of the injured persons; the first prediction network is arranged in an edge device in the accident area; A screening module, used for screening out a plurality of injured persons to be located whose injury conditions are greater than a preset threshold value from all the injured persons; The second prediction module is used to input the sensor data of the injured person to be located into a second prediction network set up in the cloud device to obtain the location information and injury information corresponding to the injured person to be located.

[0013] As an optional embodiment, in the second aspect of the present invention, the sensor data is obtained by a drone device equipped with a sensor module installed in the accident area and / or a sensor module installed on the injured person.

[0014] As an optional embodiment, in the second aspect of the present invention, the sensor data includes at least one of physiological data, motion data, environmental sensor data, image data and sound data; the physiological data includes at least one of heart rate, blood oxygen saturation, blood pressure, body temperature, respiratory rate, electrocardiogram data, blood sugar level and bleeding detection data; the motion data includes at least one of acceleration, angular velocity, displacement information and limb activity; the environmental sensor data includes at least one of temperature, humidity, air quality, dust concentration and light intensity.

[0015] As an optional embodiment, in the second aspect of the present invention, the first prediction network is trained by a training data set including multiple training casualty sensor data and corresponding injury annotations; the second prediction network is trained by a training data set including multiple training casualty sensor data and corresponding position annotations and injury annotations; the model parameter scale of the second prediction network is greater than the model parameter scale of the first prediction network.

[0016] As an optional implementation, in the second aspect of the present invention, the first prediction network and the second prediction network are trained by the following steps: Acquire a common training data set including sensor data of multiple training injured persons and corresponding injury annotations; Based on the common training data set, the preset prediction network basic model is trained until convergence to obtain a trained common prediction network model; Based on a preset distillation target parameter, distilling the common prediction network model to obtain the first prediction network; Acquire an advanced training data set including sensor data of multiple training injured persons and corresponding position annotations and injury condition annotations; The common prediction network model is trained according to the advanced training data set until convergence to obtain the second prediction network.

[0017] As an optional implementation, in the second aspect of the present invention, distilling the common prediction network model based on a preset distillation target parameter to obtain the first prediction network includes: Obtain device parameters of edge devices in the accident area; Determine the model carrying capacity parameters corresponding to the equipment parameters based on a preset parameter correspondence analysis algorithm; Inputting the model carrying capacity parameter into the model scale prediction network corresponding to the network architecture corresponding to the prediction network basic model to obtain the prediction model scale; the network architecture includes at least one of CNN architecture, RNN architecture, LSTM architecture and random forest architecture; Determining distillation target parameters according to the prediction model scale and the current model scale of the common prediction network model, and the correspondence between the preset scale and the distillation parameter; the distillation target parameters include a distillation target function value, a distillation constraint condition function, and a distillation operation duration; The common prediction network model is distilled according to the distillation target parameter to obtain the first prediction network.

[0018] As an optional implementation, in the second aspect of the present invention, the second prediction module inputs the sensor data of the injured person to be located into the second prediction network set in the cloud device to obtain the specific manner of the location information and injury information corresponding to the injured person to be located, including: For each of the injured persons to be located, the sensor data of the injured person to be located is transmitted to the cloud device, and the acquisition location and acquisition time of the sensor data and the receiving time of the cloud device are recorded; Inputting the sensor data into a second prediction network arranged in the cloud device to obtain a first predicted position and a predicted injury degree corresponding to the injured person to be located; Inputting the acquisition time and the reception time into a trained position prediction network to obtain a second predicted position; the position prediction network is trained by a training data set including a plurality of training transmission time information and corresponding position annotations; The position information corresponding to the injured person to be located is determined according to the acquired position, the first predicted position and the second predicted position.

[0019] As an optional implementation, in the second aspect of the present invention, the second prediction module determines the specific manner in which the position information corresponding to the injured person to be located is determined according to the acquired position, the first predicted position and the second predicted position, including: Calculating a position correction weight proportional to the predicted injury severity; The product of the position correction weight and the acquired position and the average value of the product, the first predicted position and the second predicted position are calculated to obtain the position information corresponding to the injured person to be located.

[0020] The third aspect of the present invention discloses another system for locating injured persons based on neural network prediction, the system comprising: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute part or all of the steps in the method for locating a wounded person based on neural network prediction disclosed in the first aspect of the present invention.

[0021] The fourth aspect of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute some or all of the steps in the method for locating a wounded person based on neural network prediction disclosed in the first aspect of the present invention.

[0022] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: Based on the sensor data of multiple injured persons in the accident area, the present invention can quickly evaluate the injury condition of each injured person through the injury prediction network in the edge device, and screen out the seriously injured injured persons to be located, and then use the positioning prediction network of the cloud device to accurately determine the location information of these injured persons, thereby achieving efficient and accurate screening and positioning of the injured, ensuring that resources are allocated first to critically injured persons in subsequent rescue, optimizing the efficiency of rescue scheduling, and improving the overall rescue effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 creative work.

[0024] Figure 1 It is a flow chart of a method for locating a wounded person based on neural network prediction disclosed in an embodiment of the present invention.

[0025] Figure 2 It is a structural schematic diagram of a wounded person positioning system based on neural network prediction disclosed in an embodiment of the present invention.

[0026] Figure 3 It is a structural schematic diagram of another wounded person positioning system based on neural network prediction disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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.

[0028] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or equipment.

[0029] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0030] The present invention discloses a method and system for locating injured persons based on neural network prediction, which can quickly evaluate the injury of each injured person based on the sensor data of multiple injured persons in the accident area through the injury prediction network in the edge device, and screen out the seriously injured injured persons to be located, and then use the positioning prediction network of the cloud device to accurately determine the location information of these injured persons, thereby achieving efficient and accurate screening and positioning of the injured, ensuring that resources are preferentially allocated to the critically injured in the subsequent rescue, optimizing the efficiency of rescue scheduling, and improving the overall rescue effect. The following are detailed descriptions.

[0031] Embodiment 1 See also Figure 1 , Figure 1 : is a flow chart of a method for locating a wounded person based on neural network prediction disclosed in an embodiment of the present invention. Figure 1 The described method for locating a casualty based on neural network prediction can be applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). Figure 1 As shown, the method for locating the injured based on neural network prediction may include the following operations: 101. Obtain sensor data of multiple injured persons in the accident area.

[0032] 102. Input the sensor data into the first prediction network to obtain the injury condition of each injured person. Optionally, the first prediction network is set in an edge device in the accident area.

[0033] 103. A plurality of injured persons to be located whose injury conditions are greater than a preset threshold are selected from all injured persons. 104. Input the sensor data of the injured person to be located into a second prediction network set up in the cloud device to obtain the location information and injury information corresponding to the injured person to be located.

[0034] It can be seen that the above-mentioned embodiments of the invention can quickly evaluate the injury condition of each injured person based on the sensor data of multiple injured persons in the accident area through the injury prediction network in the edge device, and screen out the seriously injured injured persons to be located, and then use the positioning prediction network of the cloud device to accurately determine the location information of these injured persons, thereby achieving efficient and accurate screening and positioning of the injured, ensuring that resources are allocated first to critically injured persons in subsequent rescue, optimizing the efficiency of rescue scheduling, and improving the overall rescue effect.

[0035] As an optional embodiment, in the above steps, the sensor data is obtained by a drone device equipped with a sensor module installed in the accident area and / or a sensor module installed on the injured person.

[0036] It can be seen that through the above-mentioned optional embodiments, sensor data is collected by drone equipment equipped with sensor modules in the accident area or sensor modules worn by the injured to achieve comprehensive data perception of the accident scene, improve the coverage of injury monitoring and the real-time nature of data acquisition, provide more accurate data support for subsequent injury analysis and rescue decisions, assist in achieving efficient and accurate screening and positioning of the injured, ensure that resources are allocated first to critically injured persons in subsequent rescue, optimize rescue scheduling efficiency, and improve overall rescue effects.

[0037] As an optional embodiment, in the above steps, the sensor data includes at least one of physiological data, motion data, environmental sensor data, image data and sound data; the physiological data includes at least one of heart rate, blood oxygen saturation, blood pressure, body temperature, respiratory rate, electrocardiogram data, blood sugar level and bleeding detection data; the motion data includes at least one of acceleration, angular velocity, displacement information and limb activity; the environmental sensor data includes at least one of temperature, humidity, air quality, dust concentration and light intensity.

[0038] It can be seen that through the above optional embodiments, the content of the sensor data is limited to comprehensively perceive the status of the injured and the surrounding environment in the accident area, so as to facilitate the subsequent efficient and accurate screening and positioning of the injured, ensure that resources are allocated first to critically injured people in subsequent rescue, optimize the efficiency of rescue scheduling, and improve the overall rescue effect.

[0039] As an optional embodiment, in the above steps, the first prediction network is trained by a training data set including multiple training casualty sensor data and corresponding injury annotations; the second prediction network is trained by a training data set including multiple training casualty sensor data and corresponding position annotations and injury annotations; the model parameter scale of the second prediction network is greater than the model parameter scale of the first prediction network.

[0040] It can be seen that through the above optional embodiments, the training details and model parameter details of the first prediction network and the second prediction network are limited, so that the simpler first prediction network set on the edge device can quickly realize the identification of the injuries of the wounded, while the second prediction network needs to process more complex spatial information and injury analysis, and its model parameter scale is larger than the model parameter scale of the first prediction network, thereby improving the comprehensive prediction ability of the status and location information of the wounded, assisting in the efficient and accurate screening and positioning of the wounded, ensuring that resources are allocated to critically injured people in subsequent rescue, optimizing the rescue scheduling efficiency, and improving the overall rescue effect.

[0041] As an optional embodiment, in the above steps, the first prediction network and the second prediction network are trained by the following steps: Acquire a common training data set including sensor data of multiple training injured persons and corresponding injury annotations; Based on the common training data set, the preset prediction network basic model is trained until convergence to obtain a trained common prediction network model; Based on the preset distillation target parameters, the common prediction network model is distilled to obtain a first prediction network; Acquire an advanced training data set including sensor data of multiple training injured persons and corresponding position annotations and injury condition annotations; The common prediction network model is trained according to the advanced training data set until convergence to obtain a second prediction network.

[0042] It can be seen that through the above optional embodiments, the technical details of training two prediction networks based on joint training and separate distillation optimization and training optimization steps are defined, so that the two networks can respectively achieve more reasonable and accurate recognition, assist in achieving efficient and accurate screening and positioning of the wounded, ensure that resources are allocated first to critically injured patients in subsequent rescue, optimize rescue scheduling efficiency, and improve the overall rescue effect.

[0043] As an optional embodiment, in the above steps, based on a preset distillation target parameter, the common prediction network model is distilled to obtain a first prediction network, including: Obtain device parameters of edge devices in the accident area; Based on the preset parameter correspondence analysis algorithm, determine the model carrying capacity parameters corresponding to the equipment parameters; Input the model carrying capacity parameter into the model scale prediction network corresponding to the network architecture corresponding to the prediction network basic model to obtain the prediction model scale; optionally, the network architecture includes at least one of a CNN architecture, a RNN architecture, a LSTM architecture, and a random forest architecture; Determine the distillation target parameter according to the prediction model scale and the current model scale of the common prediction network model, and the correspondence between the preset scale and the distillation parameter; optionally, the distillation target parameter includes a distillation target function value, a distillation constraint condition function, and a distillation operation time; According to the distillation target parameters, the common prediction network model is distilled to obtain the first prediction network.

[0044] It can be seen that through the above optional embodiments, based on the edge device parameters of the accident area, the parameter correspondence analysis algorithm is used to determine the model carrying capacity parameters of the equipment, and the adapted prediction model scale is inferred through the model scale prediction network, so as to reasonably determine the distillation target parameters according to the scale of the current common prediction network model and the preset scale-distillation parameter correspondence, and perform distillation optimization on the common prediction network model under the constraints of the distillation objective function, distillation constraints and distillation operation time, and obtain the first prediction network adapted to the edge device, so as to improve the deployment efficiency and reasoning performance of the prediction model in a resource-constrained environment, assist in the efficient and accurate screening and positioning of the wounded, ensure that resources are allocated to critically injured persons in subsequent rescue, optimize the rescue scheduling efficiency, and improve the overall rescue effect.

[0045] As an optional embodiment, in the above steps, the sensor data of the injured person to be located is input into a second prediction network set in a cloud device to obtain the location information and injury information corresponding to the injured person to be located, including: For each injured person to be located, the sensor data of the injured person to be located is transmitted to the cloud device, and the acquisition location and acquisition time of the sensor data and the receiving time of the cloud device are recorded; Inputting the sensor data into a second prediction network arranged in a cloud device to obtain a first predicted position and a predicted injury degree corresponding to the injured person to be located; Inputting the acquisition time and the reception time into a trained position prediction network to obtain a second predicted position; optionally, the position prediction network is trained by a training data set including a plurality of training transmission time information and corresponding position annotations; The position information corresponding to the injured person to be located is determined according to the acquired position, the first predicted position and the second predicted position.

[0046] It can be seen that through the above-mentioned optional embodiments, for the acquisition position, acquisition time and reception time of the sensor data of each injured person to be located, the sensor data is analyzed through the second prediction network of the cloud device to obtain the first predicted position and the predicted injury degree, and at the same time, the acquisition position, the first predicted position and the second predicted position are combined to comprehensively determine the position information of the injured person to be located, thereby improving the accuracy and real-time positioning of the injured, achieving efficient and accurate screening and positioning of the injured, ensuring that resources are allocated to critically injured people in subsequent rescue, optimizing the efficiency of rescue scheduling, and improving the overall rescue effect.

[0047] As an optional embodiment, in the above steps, determining the position information corresponding to the injured person to be located according to the acquired position, the first predicted position and the second predicted position includes: Calculate a position correction weight proportional to the predicted injury severity; The product of the position correction weight and the acquired position, and the average value between the product, the first predicted position and the second predicted position are calculated to obtain the position information corresponding to the injured person to be located.

[0048] It can be seen that through the above-mentioned optional embodiments, the correction weight proportional to the injury situation is limited, so that the weight of the initial sensing of the seriously injured is increased. This is because the possibility of movement of such injured persons is reduced. The position determined based on this comprehensive method is more accurate, and efficient and accurate screening and positioning of the injured are achieved, ensuring that resources are allocated first to critically injured persons in subsequent rescue operations, optimizing rescue scheduling efficiency, and improving the overall rescue effect.

[0049] Embodiment 2 See also Figure 2 , Figure 2 : is a schematic diagram of the structure of a casualty positioning system based on neural network prediction disclosed in an embodiment of the present invention. Figure 2 The described injured person positioning system based on neural network prediction can be applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). Figure 2 As shown, the casualty positioning system based on neural network prediction may include: The acquisition module 201 is used to acquire sensor data of multiple injured persons in the accident area.

[0050] The first prediction module 202 is used to input the sensor data into the first prediction network to obtain the injury condition of each injured person. Optionally, the first prediction network is set in an edge device in the accident area.

[0051] The screening module 203 is used to screen out a plurality of injured persons to be located whose injury conditions are greater than a preset threshold from all the injured persons. The second prediction module 204 is used to input the sensor data of the injured person to be located into a second prediction network set up in the cloud device to obtain the location information and injury information corresponding to the injured person to be located.

[0052] It can be seen that the above-mentioned embodiments of the invention can quickly evaluate the injury condition of each injured person based on the sensor data of multiple injured persons in the accident area through the injury prediction network in the edge device, and screen out the seriously injured injured persons to be located, and then use the positioning prediction network of the cloud device to accurately determine the location information of these injured persons, thereby achieving efficient and accurate screening and positioning of the injured, ensuring that resources are allocated first to critically injured persons in subsequent rescue, optimizing the efficiency of rescue scheduling, and improving the overall rescue effect.

[0053] As an optional embodiment, the sensor data is obtained by a drone device equipped with a sensor module installed in the accident area and / or a sensor module installed on the injured person.

[0054] It can be seen that through the above-mentioned optional embodiments, sensor data is collected by drone equipment equipped with sensor modules in the accident area or sensor modules worn by the injured to achieve comprehensive data perception of the accident scene, improve the coverage of injury monitoring and the real-time nature of data acquisition, provide more accurate data support for subsequent injury analysis and rescue decisions, assist in achieving efficient and accurate screening and positioning of the injured, ensure that resources are allocated first to critically injured persons in subsequent rescue, optimize rescue scheduling efficiency, and improve overall rescue effects.

[0055] As an optional embodiment, the sensor data includes at least one of physiological data, motion data, environmental sensor data, image data and sound data; the physiological data includes at least one of heart rate, blood oxygen saturation, blood pressure, body temperature, respiratory rate, electrocardiogram data, blood sugar level and bleeding detection data; the motion data includes at least one of acceleration, angular velocity, displacement information and limb activity; the environmental sensor data includes at least one of temperature, humidity, air quality, dust concentration and light intensity.

[0056] It can be seen that through the above optional embodiments, the content of the sensor data is limited to comprehensively perceive the status of the injured and the surrounding environment in the accident area, so as to facilitate the subsequent efficient and accurate screening and positioning of the injured, ensure that resources are allocated first to critically injured people in subsequent rescue, optimize the efficiency of rescue scheduling, and improve the overall rescue effect.

[0057] As an optional embodiment, the first prediction network is trained by a training data set including multiple training casualty sensor data and corresponding injury annotations; the second prediction network is trained by a training data set including multiple training casualty sensor data and corresponding position annotations and injury annotations; the model parameter scale of the second prediction network is greater than the model parameter scale of the first prediction network.

[0058] It can be seen that through the above optional embodiments, the training details and model parameter details of the first prediction network and the second prediction network are limited, so that the simpler first prediction network set on the edge device can quickly realize the identification of the injuries of the wounded, while the second prediction network needs to process more complex spatial information and injury analysis, and its model parameter scale is larger than the model parameter scale of the first prediction network, thereby improving the comprehensive prediction ability of the status and location information of the wounded, assisting in the efficient and accurate screening and positioning of the wounded, ensuring that resources are allocated to critically injured people in subsequent rescue, optimizing the rescue scheduling efficiency, and improving the overall rescue effect.

[0059] As an optional embodiment, the first prediction network and the second prediction network are trained by the following steps: Acquire a common training data set including sensor data of multiple training injured persons and corresponding injury annotations; Based on the common training data set, the preset prediction network basic model is trained until convergence to obtain a trained common prediction network model; Based on the preset distillation target parameters, the common prediction network model is distilled to obtain a first prediction network; Acquire an advanced training data set including sensor data of multiple training injured persons and corresponding position annotations and injury condition annotations; The common prediction network model is trained according to the advanced training data set until convergence to obtain a second prediction network.

[0060] It can be seen that through the above optional embodiments, the technical details of training two prediction networks based on joint training and separate distillation optimization and training optimization steps are defined, so that the two networks can respectively achieve more reasonable and accurate recognition, assist in achieving efficient and accurate screening and positioning of the wounded, ensure that resources are allocated first to critically injured patients in subsequent rescue, optimize rescue scheduling efficiency, and improve the overall rescue effect.

[0061] As an optional embodiment, based on a preset distillation target parameter, the common prediction network model is distilled to obtain a first prediction network, including: Obtain device parameters of edge devices in the accident area; Based on the preset parameter correspondence analysis algorithm, determine the model carrying capacity parameters corresponding to the equipment parameters; Input the model carrying capacity parameter into the model scale prediction network corresponding to the network architecture corresponding to the prediction network basic model to obtain the prediction model scale; optionally, the network architecture includes at least one of a CNN architecture, a RNN architecture, a LSTM architecture, and a random forest architecture; Determine the distillation target parameter according to the prediction model scale and the current model scale of the common prediction network model, and the correspondence between the preset scale and the distillation parameter; optionally, the distillation target parameter includes a distillation target function value, a distillation constraint condition function, and a distillation operation time; According to the distillation target parameters, the common prediction network model is distilled to obtain the first prediction network.

[0062] It can be seen that through the above optional embodiments, based on the edge device parameters of the accident area, the parameter correspondence analysis algorithm is used to determine the model carrying capacity parameters of the equipment, and the adapted prediction model scale is inferred through the model scale prediction network, so as to reasonably determine the distillation target parameters according to the scale of the current common prediction network model and the preset scale-distillation parameter correspondence, and perform distillation optimization on the common prediction network model under the constraints of the distillation objective function, distillation constraints and distillation operation time, and obtain the first prediction network adapted to the edge device, so as to improve the deployment efficiency and reasoning performance of the prediction model in a resource-constrained environment, assist in the efficient and accurate screening and positioning of the wounded, ensure that resources are allocated to critically injured persons in subsequent rescue, optimize the rescue scheduling efficiency, and improve the overall rescue effect.

[0063] As an optional embodiment, the second prediction module inputs the sensor data of the injured person to be located into the second prediction network set in the cloud device to obtain the location information and injury information corresponding to the injured person to be located, including: For each injured person to be located, the sensor data of the injured person to be located is transmitted to the cloud device, and the acquisition location and acquisition time of the sensor data and the receiving time of the cloud device are recorded; Inputting the sensor data into a second prediction network arranged in a cloud device to obtain a first predicted position and a predicted injury degree corresponding to the injured person to be located; Inputting the acquisition time and the reception time into a trained position prediction network to obtain a second predicted position; optionally, the position prediction network is trained by a training data set including a plurality of training transmission time information and corresponding position annotations; The position information corresponding to the injured person to be located is determined according to the acquired position, the first predicted position and the second predicted position.

[0064] It can be seen that through the above-mentioned optional embodiments, for the acquisition position, acquisition time and reception time of the sensor data of each injured person to be located, the sensor data is analyzed through the second prediction network of the cloud device to obtain the first predicted position and the predicted injury degree, and at the same time, the acquisition position, the first predicted position and the second predicted position are combined to comprehensively determine the position information of the injured person to be located, thereby improving the accuracy and real-time positioning of the injured, achieving efficient and accurate screening and positioning of the injured, ensuring that resources are allocated to critically injured people in subsequent rescue, optimizing the efficiency of rescue scheduling, and improving the overall rescue effect.

[0065] As an optional embodiment, the second prediction module determines the specific manner in which the position information corresponding to the to-be-located injured person is determined according to the acquired position, the first predicted position, and the second predicted position, including: Calculate a position correction weight proportional to the predicted injury severity; The product of the position correction weight and the acquired position, and the average value between the product, the first predicted position and the second predicted position are calculated to obtain the position information corresponding to the injured person to be located.

[0066] It can be seen that through the above-mentioned optional embodiments, the correction weight proportional to the injury situation is limited, so that the weight of the initial sensing of the seriously injured is increased. This is because the possibility of movement of such injured persons is reduced. The position determined based on this comprehensive method is more accurate, and efficient and accurate screening and positioning of the injured are achieved, ensuring that resources are allocated first to critically injured persons in subsequent rescue operations, optimizing rescue scheduling efficiency, and improving the overall rescue effect.

[0067] Embodiment 3 See also Figure 3 , Figure 3 This is another wounded person positioning system based on neural network prediction disclosed in an embodiment of the present invention. Figure 3 The described injured person positioning system based on neural network prediction is applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). Figure 3 As shown, the casualty positioning system based on neural network prediction may include: A memory 301 storing executable program codes; a processor 302 coupled to the memory 301; The processor 302 calls the executable program code stored in the memory 301 to execute the steps of the method for locating a wounded person based on neural network prediction described in the first embodiment.

[0068] Embodiment 4 An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the method for locating a wounded person based on neural network prediction described in the first embodiment.

[0069] Embodiment 5 An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps of the method for locating a wounded person based on neural network prediction described in Example 1.

[0070] The above describes specific embodiments of the present specification, and other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily have to be performed in the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0071] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0072] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0073] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification may be in the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the embodiments of this specification may be in the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0074] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0075] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0076] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0077] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0078] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0079] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0080] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0081] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0082] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0083] Finally, it should be noted that the method and system for locating a wounded person based on neural network prediction disclosed in the embodiment of the present invention only discloses a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions recorded in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for locating a wounded person based on neural network prediction, characterized in that: The method comprises: Acquire sensor data of multiple casualties in the accident area; Inputting the sensor data into a first prediction network to obtain the injury status of each of the injured persons; the first prediction network is arranged in an edge device in the accident area; Screening out a plurality of injured persons to be located whose injury conditions are greater than a preset threshold value from all the injured persons; The sensor data of the injured person to be located is input into a second prediction network arranged in a cloud device to obtain the position information and injury information corresponding to the injured person to be located.

2. The method for locating a wounded person based on neural network prediction according to claim 1, characterized in that: The sensor data is obtained by a drone device equipped with a sensor module installed in the accident area and / or a sensor module installed on the injured person.

3. The method for locating a wounded person based on neural network prediction according to claim 1, characterized in that: The sensor data includes at least one of physiological data, motion data, environmental sensor data, image data and sound data; the physiological data includes at least one of heart rate, blood oxygen saturation, blood pressure, body temperature, respiratory rate, electrocardiogram data, blood sugar level and bleeding detection data; the motion data includes at least one of acceleration, angular velocity, displacement information and limb activity; the environmental sensor data includes at least one of temperature, humidity, air quality, dust concentration and light intensity.

4. The method for locating a wounded person based on neural network prediction according to claim 1, characterized in that: The first prediction network is trained by a training data set including multiple training injured persons' sensor data and corresponding injury labels; the second prediction network is trained by a training data set including multiple training injured persons' sensor data and corresponding position labels and injury labels; the model parameter scale of the second prediction network is greater than the model parameter scale of the first prediction network.

5. The method for locating a wounded person based on neural network prediction according to claim 4, characterized in that: The first prediction network and the second prediction network are trained by the following steps: Acquire a common training data set including sensor data of multiple training injured persons and corresponding injury annotations; Based on the common training data set, the preset prediction network basic model is trained until convergence to obtain a trained common prediction network model; Based on a preset distillation target parameter, distilling the common prediction network model to obtain the first prediction network; Acquire an advanced training data set including sensor data of multiple training injured persons and corresponding position annotations and injury condition annotations; The common prediction network model is trained according to the advanced training data set until convergence to obtain the second prediction network.

6. The method for locating a wounded person based on neural network prediction according to claim 5, characterized in that: The step of distilling the common prediction network model based on a preset distillation target parameter to obtain the first prediction network includes: Obtain device parameters of edge devices in the accident area; Determine the model carrying capacity parameters corresponding to the equipment parameters based on a preset parameter correspondence analysis algorithm; Inputting the model carrying capacity parameter into the model scale prediction network corresponding to the network architecture corresponding to the prediction network basic model to obtain the prediction model scale; the network architecture includes at least one of CNN architecture, RNN architecture, LSTM architecture and random forest architecture; Determining distillation target parameters according to the prediction model scale and the current model scale of the common prediction network model, and the correspondence between the preset scale and the distillation parameter; the distillation target parameters include a distillation target function value, a distillation constraint condition function, and a distillation operation duration; The common prediction network model is distilled according to the distillation target parameter to obtain the first prediction network.

7. The method for locating a wounded person based on neural network prediction according to claim 1, characterized in that: The step of inputting the sensor data of the injured person to be located into a second prediction network arranged in a cloud device to obtain the location information and injury information corresponding to the injured person to be located includes: For each of the injured persons to be located, the sensor data of the injured person to be located is transmitted to the cloud device, and the acquisition location and acquisition time of the sensor data and the receiving time of the cloud device are recorded; Inputting the sensor data into a second prediction network arranged in the cloud device to obtain a first predicted position and a predicted injury degree corresponding to the injured person to be located; Inputting the acquisition time and the reception time into a trained position prediction network to obtain a second predicted position; the position prediction network is trained by a training data set including a plurality of training transmission time information and corresponding position annotations; The position information corresponding to the injured person to be located is determined according to the acquired position, the first predicted position and the second predicted position.

8. The method for locating a wounded person based on neural network prediction according to claim 7, characterized in that: The determining, according to the acquired position, the first predicted position, and the second predicted position, position information corresponding to the injured person to be located includes: Calculating a position correction weight proportional to the predicted injury severity; The product of the position correction weight and the acquired position and the average value of the product, the first predicted position and the second predicted position are calculated to obtain the position information corresponding to the injured person to be located.

9. A casualty positioning system based on neural network prediction, characterized in that: The system comprises: An acquisition module, used to acquire sensor data of multiple injured persons in the accident area; A first prediction module, used for inputting the sensor data into a first prediction network to obtain the injury condition of each of the injured persons; the first prediction network is arranged in an edge device in the accident area; A screening module, used for screening out a plurality of injured persons to be located whose injury conditions are greater than a preset threshold value from all the injured persons; The second prediction module is used to input the sensor data of the injured person to be located into a second prediction network set up in the cloud device to obtain the location information and injury information corresponding to the injured person to be located.

10. A casualty positioning system based on neural network prediction, characterized in that: The system comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the method for locating a wounded person based on neural network prediction as described in any one of claims 1-8.

Images