Fall alarm information evaluation method and device, equipment and storage medium

By performing multimodal data analysis and artificial intelligence evaluation on fall alarm information, non-real alarms are filtered out, solving the problems of resource waste and delays in existing technologies, and improving the reliability of fall detection and the quality of life of the elderly.

CN118942210APending Publication Date: 2024-11-12ALTUMAI (SHENZHEN) CO LTD
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411168430.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-23
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

Existing fall detection methods generate a large number of false fall alarm messages, resulting in wasted resources and delays in receiving accurate alarms, which affects the quality of life and safety of the elderly.

Method used

By acquiring the initial fall alarm information of the target object and its corresponding object description data, artificial intelligence technology is used to assess its authenticity, filtering out non-authentic alarm information, and multimodal data analysis and pre-trained models are used for classification and prediction to improve the accuracy of the assessment.

Benefits of technology

It reduces the processing of false alarm information, ensures the timely processing of true alarm information, improves the reliability of fall detection and the efficiency of resource utilization, and reduces the psychological burden on the elderly.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118942210A_ABST
    Figure CN118942210A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of tumble alarm analysis, and discloses a tumble alarm information evaluation method and device, equipment and a storage medium, and the method comprises the steps: obtaining initial tumble alarm information of a target object; obtaining object description data, corresponding to the initial fall alarm information, of the target object as target data; and according to the target data, carrying out authenticity evaluation on the initial tumble alarm information to obtain an authenticity evaluation result. According to the invention, authenticity evaluation is carried out on the initial tumble alarm information, a basis is provided for screening unreal tumble alarm information, a basis is provided for reducing tumble processing resources, and real tumble alarm information can be processed in time.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and fall alarm analysis technology, and in particular to a fall alarm information evaluation method, device, equipment and storage medium. Background Art

[0002] Falling is one of the most common accidental injuries among the elderly, which may cause serious physical injuries or even endanger life. Therefore, effective fall detection methods are crucial to ensure the quality of life and safety of the elderly. However, existing fall detection methods often generate a large number of false fall alarm messages, which waste resources for handling falls and may delay the processing of real fall alarm messages. Summary of the invention

[0003] Based on this, it is necessary to address the technical problem that the fall detection method in the prior art often generates a large amount of non-real fall alarm information, and propose a fall alarm information evaluation method, device, equipment and storage medium.

[0004] In a first aspect, a method for evaluating fall alarm information is provided, the method comprising:

[0005] Obtain initial fall alarm information of the target object;

[0006] Acquire object description data of the target object and corresponding to the initial fall alarm information as target data;

[0007] According to the target data, the authenticity of the initial fall alarm information is evaluated to obtain an authenticity evaluation result.

[0008] In a second aspect, a device for evaluating fall alarm information is provided, the device comprising:

[0009] An information acquisition module, used to obtain initial fall alarm information of a target object;

[0010] a target data determination module, configured to obtain object description data of the target object and corresponding to the initial fall alarm information as target data;

[0011] The authenticity assessment module is used to perform an authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result.

[0012] In a third aspect, a computer device is provided, comprising 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 above-mentioned method for evaluating fall alarm information when executing the computer program.

[0013] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned method for evaluating fall alarm information are implemented.

[0014] The fall alarm information evaluation method, device, equipment and storage medium of the present application, wherein the method obtains the initial fall alarm information of the target object, obtains the object description data of the target object and corresponding to the initial fall alarm information as the target data, and performs an authenticity evaluation on the initial fall alarm information according to the target data to obtain an authenticity evaluation result. By performing an authenticity evaluation on the initial fall alarm information, a basis is provided for screening out non-authentic fall alarm information, thereby providing a basis for reducing the resources for processing falls, and is also conducive to timely processing of authentic fall alarm information. In other words, the present application improves the reliability of fall detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] 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.

[0016] in:

[0017] Figure 1 A diagram showing an application environment of a method for evaluating fall alarm information in an embodiment;

[0018] Figure 2 is a flow chart of a method for evaluating fall alarm information in one embodiment;

[0019] Figure 3 is a structural block diagram of a device for evaluating fall alarm information in one embodiment;

[0020] Figure 4 is a structural block diagram of a computer device in one embodiment;

[0021] Figure 5 Another structural block diagram of a computer device in one embodiment. DETAILED DESCRIPTION

[0022] 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 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.

[0023] The fall alarm information evaluation method provided by the embodiment of the present invention can be applied to Figure 1 In an application environment, the application environment includes: a client 110 and a server 120, and the client 110 communicates with the server 120 through a network.

[0024] The server 120 is configured to implement the following steps: obtaining the initial fall alarm information of the target object, obtaining the object description data of the target object and corresponding to the initial fall alarm information as the target data, and performing an authenticity evaluation on the initial fall alarm information according to the target data to obtain an authenticity evaluation result. By performing an authenticity evaluation on the initial fall alarm information, a basis is provided for screening out non-real fall alarm information, thereby reducing the resources for processing falls, and is also conducive to timely processing of real fall alarm information. In other words, the present application improves the reliability of fall detection.

[0025] Optionally, the server 120 obtains the initial fall alarm information of the target object through the client 110 .

[0026] The client 110 is configured to implement the following steps: obtain the initial fall alarm information of the target object, obtain the object description data of the target object and corresponding to the initial fall alarm information as the target data, and perform an authenticity evaluation on the initial fall alarm information according to the target data to obtain an authenticity evaluation result; the client 110 sends the authenticity evaluation result to the server 120.

[0027] Among them, the client 110 can be but is not limited to various personal computers, laptops, smart phones, tablet computers, robots, drones and portable wearable devices (such as smart glasses, virtual head displays, smart watches, smart bracelets).

[0028] The server 120 may be implemented by an independent server or a server cluster consisting of multiple servers.

[0029] The present invention is described in detail below through specific embodiments.

[0030] See also Figure 2 As shown, Figure 2A flow chart of a method for evaluating fall alarm information provided by an embodiment of the present invention includes the following steps:

[0031] S1: Obtaining the initial fall alarm information of the target object;

[0032] Specifically, the initial fall alarm information of the target object input by the user can be obtained, the initial fall alarm information of the target object can be obtained from a preset storage space, and the initial fall alarm information of the target object can also be obtained from a third-party application (for example, a high-throughput distributed publish-subscribe message system).

[0033] The target object is the person who needs to be monitored for falls.

[0034] The initial fall alarm information is the alarm information generated by the fall alarm system / method when the fall alarm system / method detects that the target object falls. The fall alarm system / method can be selected from the existing technology and will not be described in detail here.

[0035] Optionally, the initial fall alarm information includes: object identification, fall location data and fall occurrence time. The object identification can be data that uniquely identifies an object (e.g., a person) such as an object name and an object ID. The fall location data is the location data of the fall. The fall occurrence time is the time when the fall occurs.

[0036] Optionally, the initial fall alarm information may also include other data, such as image data during and / or after the fall, and / or video data during the fall.

[0037] S2: Acquire object description data of the target object and corresponding to the initial fall alarm information as target data;

[0038] Specifically, according to the object identifier and the fall occurrence time in the initial fall alarm information, the object description data of the target object is acquired, and the acquired object description data is used as the target data.

[0039] Among them, the object description data of the target object input by the user and corresponding to the initial fall alarm information can be obtained, the object description data of the target object and corresponding to the initial fall alarm information can be obtained from a preset storage space, and the object description data of the target object and corresponding to the initial fall alarm information can be obtained from a third-party application.

[0040] The target data includes but is not limited to image data, video data, audio data, text, and millimeter wave detection data.

[0041] The object description data is the description data of the behavior, environment, and face of the target object within a period of time corresponding to the data collection time corresponding to the initial fall alarm information, wherein the data collection time is within the period of time.

[0042] S3: According to the target data, the authenticity of the initial fall alarm information is evaluated to obtain an authenticity evaluation result.

[0043] Specifically, based on artificial intelligence technology, the authenticity of the initial fall alarm information is evaluated according to the target data, and the evaluation result is used as the authenticity evaluation result.

[0044] The value of the authenticity evaluation result is true or false.

[0045] Optionally, the target data is input into a preset target model for authenticity classification prediction, and the classification category corresponding to the vector element with the largest value in the predicted vector is used as the authenticity evaluation result.

[0046] The target model is a pre-trained binary classification model. The model structure and model training method of the target model can be selected from the existing technology and will not be described here.

[0047] Frequent false alarms may cause tension in the elderly and their guardians, and may also cause anxiety and depression in the long run. Multiple false alarms will make users doubt the reliability of the fall alarm system / method, thereby reducing the utilization rate of the fall alarm system / method, leading to improper allocation of medical resources and social resources, and wasting precious rescue forces due to false alarms. This application conducts an authenticity assessment on the initial fall alarm information to effectively reduce the non-real fall alarm information (that is, the false initial fall alarm information) to enter the next link (that is, fall processing), which not only reduces the psychological burden of the elderly and their guardians, but also provides a basis for the emergency rescue team to provide more accurate fall alarm information, thereby ensuring the reasonable allocation of limited rescue resources. This is of great significance to improving the quality of life and social welfare of the elderly.

[0048] This embodiment obtains the initial fall alarm information of the target object, obtains the object description data of the target object and corresponding to the initial fall alarm information as the target data, and performs an authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result. By performing an authenticity assessment on the initial fall alarm information, a basis is provided for screening out non-authentic fall alarm information, thereby reducing the resources for processing falls and facilitating timely processing of authentic fall alarm information. In other words, this application improves the reliability of fall detection.

[0049] In one embodiment, the step of performing authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result includes:

[0050] S311: Determine each single indicator evaluation value according to the target data;

[0051] Specifically, according to the target data, a single indicator evaluation value is evaluated for each indicator.

[0052] S312: Determine the authenticity evaluation result according to each of the single indicator evaluation values.

[0053] Specifically, each of the single indicator evaluation values ​​is concatenated and input into a pre-trained indicator evaluation model for classification prediction, and the classification category corresponding to the vector element with the largest value in the predicted vector is taken as the authenticity evaluation result.

[0054] The indicator evaluation model is a pre-trained binary classification model. The model structure and model training method of the indicator evaluation model can be selected from the existing technology and will not be described here.

[0055] This embodiment first determines each single indicator evaluation value according to the target data, and then determines the authenticity evaluation result according to each single indicator evaluation value, thereby realizing authenticity evaluation of the initial fall alarm information based on multimodal data, providing a basis for screening out non-real fall alarm information, and further providing a basis for reducing the waste of resources for processing falls, and is also conducive to timely processing of real fall alarm information. In other words, this application improves the reliability of fall detection.

[0056] In one embodiment, the step of determining each single indicator evaluation value according to the target data includes:

[0057] S3111: determining the single index evaluation value corresponding to the position change index according to the object position data in the target data; and / or,

[0058] S3112: determining the single index evaluation value corresponding to the posture change index according to the object posture data in the target data; and / or,

[0059] S3113: Determine the single indicator evaluation value corresponding to the fall process indicator according to the fall process data in the target data; and / or,

[0060] S3114: determining the single index evaluation value corresponding to the sound index according to the sound data in the target data; and / or,

[0061] S3115: determining the single index evaluation value corresponding to the heart rate index according to the heart rate data in the target data; and / or,

[0062] S3116: Determine the single indicator evaluation value corresponding to the user feedback indicator according to the user feedback data in the target data; and / or,

[0063] S3117: Determine the single indicator evaluation value corresponding to the social network indicator according to the social network feedback data in the target data; and / or,

[0064] S3118: Determine the single index evaluation value corresponding to the facial expression index based on the facial data in the target data.

[0065] For S3111, after a fall event is detected and an alarm is triggered, the position change of the target object is monitored within a preset first time window (such as 5 seconds). If the position of the target object changes significantly during this period (for example, moves more than half a meter or one meter), the alarm is considered to be a false alarm (not real) and no alarm signal needs to be generated. In other words, a person who really falls is unlikely to move quickly in a short period of time.

[0066] Therefore, if it is determined based on the object position data in the target data that the position of the target object has changed significantly after the fall, then the single indicator evaluation value corresponding to the position change index is determined to be invalid (that is, from the perspective of the position change, the possibility of falling is relatively small); if it is determined based on the object position data in the target data that the position of the target object has not changed significantly after the fall, then the single indicator evaluation value corresponding to the position change index is determined to be valid (that is, from the perspective of the position change, the possibility of falling is relatively large).

[0067] For S3112, when the target object is identified as being in a falling state, the posture change of the target object is continuously monitored within a set second time window (e.g., 3 to 5 seconds). If the posture of the target object changes from a falling state to a standing state or from a falling state to a sitting state within the second time window, the alarm is determined to be a false alarm. According to common sense, a target object that has really fallen usually takes a certain amount of time to recover to other postures.

[0068] Therefore, if it is determined based on the object posture data in the target data that the posture change of the target object after falling is changed from a falling state to a standing state or from a falling state to a sitting state within the second time window, then the single indicator evaluation value corresponding to the posture change index is determined to be invalid (that is, from the perspective of posture change, the possibility of falling is relatively small); if it is determined based on the object posture data in the target data that the posture change of the target object after falling is not changed from a falling state to a standing state or from a falling state to a sitting state within the second time window, then the single indicator evaluation value corresponding to the posture change index is determined to be valid (that is, from the perspective of posture change, the possibility of falling is relatively large).

[0069] For S3113, the authenticity of the fall is determined by analyzing the order in which the various parts of the body touch the ground during the fall. This application mainly determines the authenticity of the fall based on unconventional falling postures. Unconventional falling postures include: head first, buttocks first, knees first, palms first, face first, side first, back first, and shoulders first.

[0070] Head first: This fall usually occurs when a person slips or loses balance, with the head hitting the ground directly. It should be noted that head first landing can cause serious brain damage, so this position is usually considered a true fall event. Hips first: When a person loses balance and falls forward or backward, the buttocks often touch the ground first. This position is common in the elderly or people with limited mobility and can cause serious injuries such as hip fractures. Knees first: In activities such as skiing or skating, knees first is a common fall position. This position can cause knee injuries, especially for people who already have joint problems. Palms first: When trying to stop a fall, people often put their hands out to support their bodies. Palms first landing can reduce the impact on other parts of the body, but it can also cause wrist or arm injuries. Face first landing: When a person falls forward and fails to use their arms to support themselves in time, the face may hit the ground directly. This position may cause injuries to the nose, teeth or eyes. Side first: When a person falls, if the body leans to one side, this position may cause one side of the body to land first. This position may cause rib fractures or other internal organ damage. Back first: When a person loses balance and falls backward, the back may hit the ground directly. This position may cause spinal damage or other back injuries. Shoulder first: When a person falls, the shoulder may land first, especially when trying to twist the body to reduce the injury to other parts of the body. This position may cause a shoulder dislocation or shoulder fracture.

[0071] Therefore, according to the falling process data in the target data, it is determined whether the change of the landing part of the target object during the falling process meets the requirements of the change of the landing part during the falling process. If not, the single indicator evaluation value corresponding to the falling process indicator is determined to be invalid (that is, from the change of the landing part during the falling process, the possibility of falling is relatively small). If it meets the requirements, the single indicator evaluation value corresponding to the falling process indicator is determined to be valid (that is, from the change of the landing part during the falling process, the possibility of falling is relatively large).

[0072] For S3114, analyze the sound changes before and after the fall, such as whether there are screams, cries for help or other abnormal sounds. Use audio analysis technology to determine whether someone has issued a distress signal. Audio analysis technology relies on a built-in microphone or an external audio input device to monitor sound changes in the environment. When a person falls, it is usually accompanied by screams, cries for help or other abnormal sounds. By analyzing the changes in these sounds, it can be determined whether a fall has occurred. Optionally, when the sound change is detected to exceed a preset sound change threshold, it can be determined whether a fall has occurred. The preset sound change threshold should be determined based on pre-collected baseline data of sound changes and experiments to distinguish between real fall events and daily activities. Optionally, sound change analysis is used to analyze sound change trends and sound intensity. A fall is usually accompanied by a sudden change in sound intensity, and a specific algorithm is used to assess the possibility of a fall.

[0073] Therefore, sound features are extracted based on the sound data in the target data, the sound features are input into the pre-trained sound prediction model for classification prediction, and the classification category corresponding to the vector element with the largest value in the predicted vector is used as the single indicator evaluation value corresponding to the sound indicator. The single indicator evaluation value corresponding to the sound indicator is valid or invalid.

[0074] The sound prediction model is a pre-trained binary classification model. The model structure and model training method based on the sound prediction model can be selected from the existing technology and will not be described in detail here.

[0075] For the S3115, real falls are often accompanied by sudden changes in heart rate. Heart rate monitoring technology relies on built-in cameras to capture tiny color changes in parts such as the face or wrist, which reflect changes in blood flow. Through the video heart rate algorithm, heart rate information can be extracted from video data. When a person falls, it is usually accompanied by a sudden change in heart rate, which can be used to determine whether a fall has occurred.

[0076] Fall detection, that is, when the heart rate change is detected to exceed the preset heart rate change threshold, it can be determined whether a fall event has occurred; the preset heart rate change threshold should be determined based on the pre-collected baseline data of heart rate changes and experiments to distinguish between real fall events and daily activities.

[0077] Heart rate change analysis: Analyze the trend of heart rate changes and the intensity of heart rate. Use a specific algorithm based on heart rate change analysis to assess the possibility of a fall; heart rate change, that is, a fall is usually accompanied by a sudden change in heart rate; heart rate peak, that is, analyze the size and duration of the heart rate peak; use the video heart rate algorithm to analyze the trend of heart rate changes. For example, if the heart rate suddenly increases or decreases, and the change exceeds the preset heart rate change threshold, it is considered a real fall event.

[0078] Threshold application: If a potential fall event lasts for more than a set time (such as 1 second) and the heart rate change meets the preset conditions, it is considered a real fall event.

[0079] Therefore, based on the feature extraction model trained by the neural network, the heart rate features are extracted from the heart rate data in the target data, the heart rate features are input into the pre-trained heart rate prediction model for classification prediction, and the classification category corresponding to the vector element with the largest value in the predicted vector is used as the single indicator evaluation value corresponding to the heart rate indicator. The heart rate prediction model is a pre-trained two-classification model. The model structure and model training method based on the heart rate prediction model can be selected from the existing technology and will not be elaborated here.

[0080] For S3116, the authenticity of the initial fall alarm information can be manually confirmed by the user. Design a simple and quick way for users to confirm whether they are safe, thereby reducing false alarms. User feedback technology relies on users to provide immediate confirmation information to verify the effectiveness of the alarm. By designing a simple and quick feedback mechanism, users can confirm whether they really need help, thereby reducing false alarms.

[0081] Alarm triggering: When the system detects a fall event and generates an initial fall alarm message, a confirmation request is sent to the user.

[0082] User feedback: The user provides feedback through the interactive device based on the confirmation request, such as pressing a button, speaking a specific voice command, a specific body movement, a gesture (such as raising a hand for three seconds), or one or more of these to confirm whether the target object really needs help.

[0083] Feedback processing: If the user feedback confirms that a fall event has occurred, the alarm process continues. If the user feedback confirms that no fall event has occurred, there is no need to generate an alarm signal.

[0084] False alarm prevention: If no response is received within a set time, it is assumed that help is needed.

[0085] Therefore, if the user feedback data in the target data is that help is needed, the single indicator evaluation value corresponding to the user feedback indicator is determined to be valid; if the user feedback data in the target data is that help is not needed, the single indicator evaluation value corresponding to the user feedback indicator is determined to be invalid.

[0086] For S3117, when the authenticity of the initial fall alarm information is uncertain, you can try to contact the user's emergency contacts or friends in the social network for confirmation. Social network contact technology relies on contacting the user's emergency contacts or friends in the social network to confirm the authenticity of the initial fall alarm information; when it is not possible to determine whether the initial fall alarm information is true, you can try to contact these pre-set contacts to obtain more information; allow users to pre-set emergency contact information, which includes name, phone number, email address, etc.; users can also choose whether to authorize access to their social network accounts to contact friends.

[0087] Uncertain alarm processing: When a fall event is detected and an initial fall alarm message is generated but the authenticity of the initial fall alarm message cannot be determined, the uncertain alarm processing procedure is started; if the user does not provide feedback in time (user feedback data in step S3116), an attempt is made to contact an emergency contact or friends in the social network.

[0088] Contact emergency contact: Send an emergency contact request to the emergency contact, explain the situation and request confirmation. The request should include the user's current location and possible cause of the fall.

[0089] Contact social network friends: If the user authorizes access to their social network account, they can contact their friends in the social network and send an emergency contact request to the selected friends, explaining the situation and requesting confirmation.

[0090] Waiting for reply: Wait for a reply from the emergency contact or friend. If a reply is received within the set time, decide whether to proceed to the next step (i.e. fall handling) based on the content of the reply.

[0091] Feedback processing: If the emergency contact or friend confirms that a fall has occurred, the alarm process continues.

[0092] False alarm prevention: If no response is received within a set time, it is assumed that help is needed.

[0093] Therefore, if the social network feedback data in the target data includes replies that require help, the single indicator evaluation value corresponding to the social network indicator is determined to be valid; if the social network feedback data in the target data does not include replies that require help, the single indicator evaluation value corresponding to the social network indicator is determined to be invalid.

[0094] For S3118, the facial data in the target data is input into a pre-trained facial expression prediction model for classification prediction, and the classification category corresponding to the vector element with the largest value in the predicted vector is used as the single indicator evaluation value corresponding to the facial expression indicator. The facial expression prediction model is a pre-trained binary classification model. The model structure and model training method based on the facial expression prediction model can be selected from the existing technology and will not be described in detail here.

[0095] The target data described in this embodiment evaluates the single indicator evaluation value for each indicator, providing a basis for implementing authenticity evaluation of the initial fall alarm information based on multi-modal data.

[0096] In one embodiment, the step of determining the single index evaluation value corresponding to the position change index according to the object position data in the target data includes:

[0097] S31111: using a first time window and the generation time of the initial fall alarm information as the start time of the first time window, extracting data from the object position data to obtain first data;

[0098] Specifically, the generation time of the initial fall alarm information is used as the start time of the first time window, and the first time window is used as the longest extraction duration, data is extracted from the object position data, and the extracted data is used as the first data. In other words, the first data is the descriptive data for determining the position of the target object after the fall corresponding to the initial fall alarm information.

[0099] The object position data may be determined based on one or more of image data, video data, and millimeter wave detection data.

[0100] The object location data represents the association between the detection time and location data.

[0101] S31112: Determine, based on the first data, whether the moving distance of the target object exceeds a first threshold;

[0102] S31113: If exceeded, determining that the single indicator evaluation value corresponding to the position change indicator is invalid;

[0103] Specifically, if it exceeds, that is, the moving distance of the target object exceeds the first threshold, which means that the position of the target object has changed significantly and the possibility of falling is relatively small. Therefore, the single indicator evaluation value corresponding to the position change indicator is determined to be invalid.

[0104] S31114: If not exceeded, determining that the single indicator evaluation value corresponding to the position change indicator is valid.

[0105] Specifically, if it does not exceed, it means that the moving distance of the target object does not exceed the first threshold, which means that the position of the target object has not changed significantly and there is a high possibility of falling. Therefore, the single indicator evaluation value corresponding to the position change indicator is determined to be valid.

[0106] This embodiment uses a first time window and takes the generation time of the initial fall alarm information as the start time of the first time window to extract data from the object position data, determines whether the position of the target object has changed significantly within a period of time, and improves the accuracy of the single indicator evaluation value corresponding to the position change indicator based on the movement logic of the person after the fall.

[0107] In one embodiment, the step of determining the single index evaluation value corresponding to the posture change index according to the object posture data in the target data further includes:

[0108] S31121: using a second time window and the generation time of the initial fall alarm information as the start time of the second time window, extracting data from the object posture data to obtain second data;

[0109] Specifically, the generation time of the initial fall alarm information is used as the start time of the second time window, and the second time window is used as the longest extraction duration, data is extracted from the object posture data, and the extracted data is used as the second data. In other words, the second data is the description data of the posture of the target object after falling corresponding to the initial fall alarm information.

[0110] The object posture data may be determined based on one or more of image data, video data, and millimeter wave detection data.

[0111] The object posture data represents the association between the detection time and the posture data.

[0112] S31122: Determine posture change data of the target object according to the second data;

[0113] Specifically, the second data is input into a pre-trained posture change model for classification prediction, and the classification category (a combination of one or more postures) corresponding to the vector element with the largest value in the predicted vector is used as the posture change data of the target object.

[0114] The posture change model is a pre-trained multi-classification model. The model structure and model training method of the posture change model can be selected from the existing technology and will not be described in detail here.

[0115] S31123: Inputting the posture change data into a pre-trained first model for classification prediction to obtain a first prediction vector;

[0116] The pre-trained first model is a pre-trained multi-classification model. The model structure and model training method of the pre-trained first model can be selected from the prior art and will not be described in detail here.

[0117] S31124: Determine the single index evaluation value corresponding to the posture change index based on the first prediction vector.

[0118] Specifically, the classification category corresponding to the vector element with the largest value in the first prediction vector is used as the single index evaluation value corresponding to the posture change index.

[0119] This embodiment improves the accuracy of the determined single indicator evaluation value corresponding to the posture change index by extracting data from the object posture data by adopting a second time window and taking the generation time of the initial fall alarm information as the start time of the second time window.

[0120] In one embodiment, the step of determining a single indicator evaluation value corresponding to a falling process indicator according to the object posture data of the target data further includes:

[0121] S31131: using a third time window and the generation time of the initial fall alarm information as the end time of the third time window, extracting data from the object posture data in the target data to obtain third data;

[0122] Specifically, the generation time of the initial fall alarm information is used as the end time of the third time window, and the third time window is used as the longest extraction duration, data is extracted from the object posture data in the target data, and the extracted data is used as the third data. In other words, the third data is the description data of the fall process corresponding to the initial fall alarm information.

[0123] S31132: extracting a fall process feature according to the third data, where the fall process feature is a feature describing a body part landing on the ground;

[0124] Specifically, a feature extraction model obtained through training of a neural network model is used to extract the fall process features from the third data.

[0125] S31133: Inputting the fall process feature into a pre-trained second model for classification prediction to obtain a second prediction vector;

[0126] The pre-trained second model is a pre-trained multi-classification model. The model structure and model training method of the pre-trained second model can be selected from the prior art and will not be described in detail here.

[0127] S31134: Determine a single indicator evaluation value corresponding to the fall process indicator according to the second prediction vector.

[0128] Specifically, the classification category corresponding to the vector element with the largest value in the second prediction vector is used as the single indicator evaluation value corresponding to the fall process indicator.

[0129] This embodiment determines a single indicator evaluation value corresponding to the fall process indicator by extracting data from the object posture data in the target data by taking the generation time of the initial fall alarm information as the end time of the third time window, thereby improving the accuracy of the determined single indicator evaluation value corresponding to the fall process indicator.

[0130] In one embodiment, the step of determining each single indicator evaluation value according to the target data further includes:

[0131] S31191: determining first sensitivity data according to the area risk type and / or fall environment data corresponding to the initial fall alarm information, wherein the fall environment data includes: one or more of ambient lighting data, ground material data, and obstacle data;

[0132] The first sensitivity data may be a specific level or a specific value.

[0133] The possibility of falling is assessed by combining factors such as ambient lighting and ground material. Environmental factor technology relies on analyzing the specific environment where a fall occurs, including the identification of high-risk areas. When the monitoring area is defined as a bathroom or balcony, staircase or an area with a high risk factor of slippery ground, the detection sensitivity needs to be automatically increased.

[0134] Specifically, a table lookup method is used to determine the first sub-sensitivity data according to the regional risk type corresponding to the initial fall alarm information; the fall environment data corresponding to the initial fall alarm information is input into a pre-trained sensitivity prediction model for classification prediction, and the classification category corresponding to the vector element with the largest value in the predicted vector is used as the second sub-sensitivity data.

[0135] The sensitivity prediction model is a pre-trained multi-classification model. The model structure and model training method of the sensitivity prediction model can be selected from the existing technology.

[0136] If the first sensitivity data is determined according to the regional risk type and the fall environment data corresponding to the initial fall alarm information, the first sensitivity data and the second sub-sensitivity data are fused to obtain the first sensitivity data. For example, if the first sensitivity data and the second sub-sensitivity data are both specific values, the first sensitivity data and the second sub-sensitivity data are weighted and summed, and the data obtained by the weighted sum is used as the first sensitivity data.

[0137] If the first sensitivity data is determined according to the area risk type corresponding to the initial fall alarm information, the first sensitivity data is used as the first sensitivity data.

[0138] If the first sensitivity data is determined according to the fall environment data corresponding to the initial fall alarm information, the second sub-sensitivity data is used as the first sensitivity data.

[0139] The ambient lighting data describes the lighting of the environment in which the target object is located when the fall corresponding to the initial fall alarm information occurs.

[0140] The ground material data describes the material of the ground where the target object is located when the fall corresponding to the initial fall alarm information occurs.

[0141] The obstacle data describes the obstacles in the environment where the target object is located when the fall corresponding to the initial fall alarm information occurs.

[0142] S31192: Determine each of the single indicator evaluation values ​​based on the first sensitivity data and the target data.

[0143] Specifically, a table lookup method is used to determine the evaluation configuration data according to the first sensitivity data, and each of the single indicator evaluation values ​​is determined according to the evaluation configuration data and the target data.

[0144] This embodiment determines the first sensitivity data according to the regional risk type and / or fall environment data corresponding to the initial fall alarm information, and then uses the first sensitivity data to determine each of the single indicator evaluation values, thereby realizing determination of each of the single indicator evaluation values ​​based on the risk level of the region and / or the degree of influence of the environment on the fall, thereby improving the accuracy of the determined each of the single indicator evaluation values, and thereby improving the accuracy of the authenticity assessment.

[0145] In one embodiment, the step of determining each single indicator evaluation value according to the target data further includes:

[0146] S41: Acquire first sensitivity data and second sensitivity data input by a user;

[0147] Allowing users to adjust the sensitivity of fall detection (i.e., input second sensitivity data) based on their health status and activity habits. The adjustable sensitivity setting technology relies on the user to adjust the sensitivity of the fall detection system based on their health status and activity habits. This allows the response of the fall detection system to be customized according to individual needs, thereby reducing false alarms and improving accuracy.

[0148] User interface design: Design a clear and concise user interface so that users can easily adjust sensitivity settings. Display the current sensitivity data and provide clear instructions to guide users on how to adjust the sensitivity.

[0149] Interactive device initialization: Initialize interactive devices such as buttons, touch screens, or voice recognition modules. Set the operation logic of the interactive device to ensure that users can easily use the interactive device to input the second sensitivity data.

[0150] User Settings: Allow users to adjust sensitivity according to their health status and activity habits, and provide different sensitivity data, such as adjustment options of levels or specific values.

[0151] Sensitivity adjustment: Automatically update the fall detection threshold when the user changes the sensitivity setting. For example, increasing the sensitivity may increase the risk of false alarms, but also improve the ability to detect fall events; decreasing the sensitivity has the opposite effect.

[0152] Alarm triggering: When a fall event is detected and the initial fall alarm information is generated, the validity of the second sensitivity data is determined according to the sensitivity data set by the user (that is, the second sensitivity data).

[0153] Specifically, the user may input the second sensitivity data through the client or through a third-party application.

[0154] The second sensitivity data may be a specific level or a specific value.

[0155] S42: determining evaluation configuration data according to the first sensitivity data and the second sensitivity data, wherein the evaluation configuration data describes a threshold value required for determining the single indicator evaluation value;

[0156] Optionally, a table lookup method is used to determine the evaluation configuration data based on the first sensitivity data and the second sensitivity data.

[0157] Optionally, a weighted sum is performed on the first sensitivity data and the second sensitivity data, and a table lookup method is used to determine the evaluation configuration data based on the data obtained by the weighted sum.

[0158] S43: Determine each of the single indicator evaluation values ​​according to the evaluation configuration data and the target data.

[0159] Specifically, based on the threshold in the evaluation configuration data and according to the target data, each of the single indicator evaluation values ​​is determined.

[0160] This embodiment determines each of the single indicator evaluation values ​​based on the target data based on the threshold value in the evaluation configuration data determined according to the first sensitivity data and the second sensitivity data input by the user, thereby realizing the determination of each of the single indicator evaluation values ​​based on the risk level of the area and / or the degree of influence of the environment on the fall combined with the user's choice (second sensitivity data), improving the accuracy of the determined single indicator evaluation values, meeting the user's personalized needs, and thereby improving the accuracy of the authenticity assessment.

[0161] In one embodiment, second sensitivity data input by a user is obtained; evaluation configuration data is determined based on the second sensitivity data; and each of the single indicator evaluation values ​​is determined based on the evaluation configuration data and the target data.

[0162] In one embodiment, the target data is data obtained by performing privacy calculation based on privacy technology;

[0163] The privacy technology includes: one or more of a stickman generation technology and a grid-based 3D human body model generation technology.

[0164] Specifically, the image data in the target data is data obtained by privacy calculation based on privacy technology, and the video data in the target data is data obtained by privacy calculation based on privacy technology.

[0165] The stickman generation technology is to replace the characters in the image or video with stickman. The specific implementation method of the stickman generation technology can be selected from the existing technology and will not be described in detail here.

[0166] The grid-based 3D (3D) human body model generation technology replaces the characters in the image or video with a grid-based 3D human body model. The specific implementation method of the grid-based 3D human body model generation technology can be selected from the existing technology and will not be described in detail here.

[0167] Optionally, the mesh-based 3D human body model is a cartoon-style 3D human body model.

[0168] It is understandable that when the data is obtained through privacy calculation based on privacy technology, privacy calculation can be performed on part or all of the image area corresponding to the person.

[0169] The target data of this embodiment is data obtained by privacy calculation based on privacy technology, making this application suitable for fall detection scenarios that need to protect the privacy of people.

[0170] In one embodiment, after the step of performing authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result, the method further includes:

[0171] S51: If the authenticity evaluation result is true, adding the initial fall alarm information and the target data to a real case library, and generating an alarm signal according to the initial fall alarm information;

[0172] Specifically, if the authenticity assessment result is true, it means that the initial fall alarm information is true. Therefore, an alarm signal is generated according to the initial fall alarm information, and the next link (that is, fall processing) is started through the alarm signal.

[0173] S52: If the authenticity evaluation result is untrue, adding the initial fall alarm information and the target data to an untrue case library;

[0174] The real case library and the unreal case library are used to update the algorithm and / or model used in the step of performing authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result.

[0175] Specifically, if the authenticity assessment result is true, it means that the initial fall alarm information is not true and is a false alarm, and there is no need to start the next link (ie, fall processing).

[0176] The real case library and the unreal case library are used to update the coefficients of the algorithm and / or the network parameters of the model used in the step of performing authenticity assessment on the initial fall alarm information according to the target data to obtain the authenticity assessment result, thereby realizing self-learning and providing a basis for further improving the accuracy of authenticity assessment.

[0177] In this embodiment, when the authenticity evaluation result is true, an alarm signal is generated according to the initial fall alarm information. When the authenticity evaluation result is untrue, there is no need to generate an alarm signal according to the initial fall alarm information, so that the true initial fall alarm information enters the next link (that is, fall processing), reducing the resources for processing falls, which is conducive to timely processing of true fall alarm information.

[0178] See also Figure 3 As shown, in one embodiment, a device for evaluating fall alarm information is provided, the device comprising:

[0179] An information acquisition module 801 is used to acquire initial fall alarm information of a target object;

[0180] A target data determination module 802 is used to obtain object description data of the target object and corresponding to the initial fall alarm information as target data;

[0181] The authenticity assessment module 803 is used to perform an authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result.

[0182] This embodiment obtains the initial fall alarm information of the target object, obtains the object description data of the target object and corresponding to the initial fall alarm information as the target data, and performs an authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result. By performing an authenticity assessment on the initial fall alarm information, a basis is provided for screening out non-authentic fall alarm information, thereby reducing the resources for processing falls and facilitating timely processing of authentic fall alarm information. In other words, this application improves the reliability of fall detection.

[0183] In one embodiment, the step of performing authenticity evaluation on the initial fall alarm information according to the target data to obtain an authenticity evaluation result in the authenticity evaluation module 803 includes:

[0184] Determine the evaluation value of each single indicator according to the target data;

[0185] The authenticity evaluation result is determined according to each of the single indicator evaluation values.

[0186] In one embodiment, the step of determining each single indicator evaluation value according to the target data in the authenticity evaluation module 803 includes:

[0187] Determining the single index evaluation value corresponding to the position change index according to the object position data in the target data; and / or,

[0188] Determining the single index evaluation value corresponding to the posture change index according to the object posture data in the target data; and / or,

[0189] Determining the single indicator evaluation value corresponding to the falling process indicator according to the falling process data in the target data; and / or,

[0190] Determining the single index evaluation value corresponding to the sound index according to the sound data in the target data; and / or,

[0191] Determining the single index evaluation value corresponding to the heart rate index according to the heart rate data in the target data; and / or,

[0192] Determining the single indicator evaluation value corresponding to the user feedback indicator according to the user feedback data in the target data; and / or,

[0193] Determining the single indicator evaluation value corresponding to the social network indicator according to the social network feedback data in the target data; and / or,

[0194] According to the facial data in the target data, the single index evaluation value corresponding to the facial expression index is determined.

[0195] In one embodiment, the step of determining the single index evaluation value corresponding to the position change index according to the object position data in the target data in the authenticity evaluation module 803 includes:

[0196] Using a first time window and the generation time of the initial fall alarm information as the start time of the first time window, extracting data from the object position data to obtain first data;

[0197] Determining, based on the first data, whether the moving distance of the target object exceeds a first threshold;

[0198] If it exceeds, the single indicator evaluation value corresponding to the position change indicator is determined to be invalid;

[0199] If not, the single indicator evaluation value corresponding to the position change indicator is determined to be valid.

[0200] In one embodiment, the step of determining the single index evaluation value corresponding to the posture change index according to the object posture data in the target data in the authenticity evaluation module 803 further includes:

[0201] Using a second time window and the generation time of the initial fall alarm information as the start time of the second time window, extracting data from the object posture data to obtain second data;

[0202] determining posture change data of the target object according to the second data;

[0203] Inputting the posture change data into a pre-trained first model for classification prediction to obtain a first prediction vector;

[0204] According to the first prediction vector, the single index evaluation value corresponding to the posture change index is determined.

[0205] In one embodiment, the step of determining the single indicator evaluation value corresponding to the fall process indicator according to the object posture data of the target data in the authenticity evaluation module 803 further includes:

[0206] Using a third time window and the generation time of the initial fall alarm information as the end time of the third time window, extracting data from the object posture data in the target data to obtain third data;

[0207] extracting a falling process feature according to the third data, wherein the falling process feature is a feature describing a body part landing on the ground;

[0208] Inputting the fall process feature into a pre-trained second model for classification prediction to obtain a second prediction vector;

[0209] According to the second prediction vector, a single indicator evaluation value corresponding to the fall process indicator is determined.

[0210] In one embodiment, the step of determining each single indicator evaluation value according to the target data in the authenticity evaluation module 803 further includes:

[0211] Determine first sensitivity data according to the area risk type and / or fall environment data corresponding to the initial fall alarm information, wherein the fall environment data includes: one or more of ambient lighting data, ground material data, and obstacle data;

[0212] Determine each of the single indicator evaluation values ​​according to the first sensitivity data and the target data.

[0213] In one embodiment, the step of determining each single indicator evaluation value according to the target data in the authenticity evaluation module 803 further includes:

[0214] Acquire first sensitivity data and second sensitivity data input by a user;

[0215] Determine evaluation configuration data according to the first sensitivity data and the second sensitivity data, wherein the evaluation configuration data expresses a threshold value required for determining the single indicator evaluation value;

[0216] Determine each of the single indicator evaluation values ​​according to the evaluation configuration data and the target data.

[0217] In one embodiment, the target data is data obtained by performing privacy calculation based on privacy technology;

[0218] The privacy technology includes: one or more of a stickman generation technology and a grid-based 3D human body model generation technology.

[0219] In one embodiment, the device further includes: an application module, wherein the application module is configured to:

[0220] If the authenticity evaluation result is true, adding the initial fall alarm information and the target data to a real case library, and generating an alarm signal according to the initial fall alarm information;

[0221] If the authenticity evaluation result is untrue, adding the initial fall alarm information and the target data to an untrue case library;

[0222] The real case library and the unreal case library are used to update the algorithm and / or model used in the step of performing authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result.

[0223] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, the functions or steps on the server side of a method for evaluating fall alarm information are implemented.

[0224] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, the functions or steps on the client side of a method for evaluating fall alarm information are implemented.

[0225] In one embodiment, a computer device is provided, the computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0226] Obtain initial fall alarm information of the target object;

[0227] Acquire object description data of the target object and corresponding to the initial fall alarm information as target data;

[0228] According to the target data, the authenticity of the initial fall alarm information is evaluated to obtain an authenticity evaluation result.

[0229] This embodiment obtains the initial fall alarm information of the target object, obtains the object description data of the target object and corresponding to the initial fall alarm information as the target data, and performs an authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result. By performing an authenticity assessment on the initial fall alarm information, a basis is provided for screening out non-authentic fall alarm information, thereby reducing the resources for processing falls and facilitating timely processing of authentic fall alarm information. In other words, this application improves the reliability of fall detection.

[0230] In one embodiment, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0231] Obtain initial fall alarm information of the target object;

[0232] Acquire object description data of the target object and corresponding to the initial fall alarm information as target data;

[0233] According to the target data, the authenticity of the initial fall alarm information is evaluated to obtain an authenticity evaluation result.

[0234] This embodiment obtains the initial fall alarm information of the target object, obtains the object description data of the target object and corresponding to the initial fall alarm information as the target data, and performs an authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result. By performing an authenticity assessment on the initial fall alarm information, a basis is provided for screening out non-authentic fall alarm information, thereby reducing the resources for processing falls and facilitating timely processing of authentic fall alarm information. In other words, this application improves the reliability of fall detection.

[0235] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can refer to the relevant descriptions on the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0236] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0237] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0238] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for evaluating fall alarm information, characterized in that: The method comprises: Obtain initial fall alarm information of the target object; Acquire object description data of the target object and corresponding to the initial fall alarm information as target data; According to the target data, the authenticity of the initial fall alarm information is evaluated to obtain an authenticity evaluation result.

2. The method for evaluating fall alarm information according to claim 1, characterized in that: The step of performing authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result comprises: Determine the evaluation value of each single indicator according to the target data; The authenticity evaluation result is determined according to each of the single indicator evaluation values.

3. The method for evaluating fall alarm information according to claim 2, characterized in that: The step of determining each single indicator evaluation value according to the target data includes: Determining the single index evaluation value corresponding to the position change index according to the object position data in the target data; and / or, Determining the single index evaluation value corresponding to the posture change index according to the object posture data in the target data; and / or, Determining the single indicator evaluation value corresponding to the falling process indicator according to the falling process data in the target data; and / or, Determining the single index evaluation value corresponding to the sound index according to the sound data in the target data; and / or, Determining the single index evaluation value corresponding to the heart rate index according to the heart rate data in the target data; and / or, Determining the single indicator evaluation value corresponding to the user feedback indicator according to the user feedback data in the target data; and / or, Determining the single indicator evaluation value corresponding to the social network indicator according to the social network feedback data in the target data; and / or, According to the facial data in the target data, the single index evaluation value corresponding to the facial expression index is determined.

4. The method for evaluating fall alarm information according to claim 3, characterized in that: The step of determining the single index evaluation value corresponding to the position change index according to the object position data in the target data comprises: Using a first time window and the generation time of the initial fall alarm information as the start time of the first time window, extracting data from the object position data to obtain first data; Determining, based on the first data, whether the moving distance of the target object exceeds a first threshold; If it exceeds, the single indicator evaluation value corresponding to the position change indicator is determined to be invalid; If not, the single indicator evaluation value corresponding to the position change indicator is determined to be valid.

5. The method for evaluating fall alarm information according to claim 3, characterized in that: The step of determining the single index evaluation value corresponding to the posture change index according to the object posture data in the target data further includes: Using a second time window and the generation time of the initial fall alarm information as the start time of the second time window, extracting data from the object posture data to obtain second data; determining posture change data of the target object according to the second data; Inputting the posture change data into a pre-trained first model for classification prediction to obtain a first prediction vector; According to the first prediction vector, the single index evaluation value corresponding to the posture change index is determined.

6. The method for evaluating fall alarm information according to claim 3, characterized in that: The step of determining a single indicator evaluation value corresponding to a falling process indicator according to the object posture data of the target data further includes: Using a third time window and the generation time of the initial fall alarm information as the end time of the third time window, extracting data from the object posture data in the target data to obtain third data; extracting a falling process feature according to the third data, wherein the falling process feature is a feature describing a body part landing on the ground; Inputting the fall process feature into a pre-trained second model for classification prediction to obtain a second prediction vector; According to the second prediction vector, a single indicator evaluation value corresponding to the fall process indicator is determined.

7. The method for evaluating fall alarm information according to claim 2, characterized in that: The step of determining each single indicator evaluation value according to the target data further includes: Determine first sensitivity data according to the area risk type and / or fall environment data corresponding to the initial fall alarm information, wherein the fall environment data includes: one or more of ambient lighting data, ground material data, and obstacle data; Determine each of the single indicator evaluation values ​​according to the first sensitivity data and the target data.

8. The method for evaluating fall alarm information according to claim 2 or 7, characterized in that: The step of determining each single indicator evaluation value according to the target data further includes: Acquire first sensitivity data and second sensitivity data input by a user; Determine evaluation configuration data according to the first sensitivity data and the second sensitivity data, wherein the evaluation configuration data expresses a threshold value required for determining the single indicator evaluation value; Determine each of the single indicator evaluation values ​​according to the evaluation configuration data and the target data.

9. The method for evaluating fall alarm information according to claim 2, characterized in that: The target data is data obtained by performing privacy calculation based on privacy technology; The privacy technology includes: one or more of a stickman generation technology and a grid-based 3D human body model generation technology.

10. The method for evaluating fall alarm information according to claim 1, characterized in that: After the step of performing authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result, the method further includes: If the authenticity evaluation result is true, adding the initial fall alarm information and the target data to a real case library, and generating an alarm signal according to the initial fall alarm information; If the authenticity evaluation result is untrue, adding the initial fall alarm information and the target data to an untrue case library; The real case library and the unreal case library are used to update the algorithm and / or model used in the step of performing authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result.

11. A device for evaluating fall alarm information, characterized in that: The device comprises: An information acquisition module, used to obtain initial fall alarm information of a target object; a target data determination module, configured to obtain object description data of the target object and corresponding to the initial fall alarm information as target data; The authenticity assessment module is used to perform an authenticity assessment on the initial fall alarm information according to the target data to obtain an authenticity assessment result.

12. A computer device, characterized in that: The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the method for evaluating fall alarm information according to any one of claims 1 to 10 are implemented.

13. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for evaluating fall warning information according to any one of claims 1 to 10 are implemented.

Citation Information

Cited By

  • Emotional recognition and behavior pattern analysis-based early warning and intervention method for falling down of old people

    CN119723805A

  • Fall warning and intervention methods for the elderly based on emotion recognition and behavioral pattern analysis

    CN119723805B