Pet abnormal behavior identification method and system

Abstract

The invention discloses a pet abnormal behavior identification method and system. The method comprises the following steps: preprocessing pet historical behavior data to obtain effective behavior data; calculating a dynamic behavior abnormal coefficient threshold based on the historical dynamic behavior data, and calculating a static behavior abnormal coefficient threshold based on the historical static behavior data; when the current behavior of the pet belongs to the dynamic behavior, executing pet abnormal behavior judgment according to a dynamic behavior abnormal coefficient threshold value and the current behavior data of the pet; when the current behavior of the pet belongs to the static behavior, executing pet abnormal behavior judgment according to a static behavior abnormal coefficient threshold value and the current behavior data of the pet; and acquiring the number of abnormal behaviors of the pet in a preset period, and when the number of abnormal behaviors of the pet reaches an abnormal number threshold, considering that the pet is abnormal and giving a prompt. The method has the beneficial effects that the abnormity of the pet in the active state and the abnormity of the pet in the static state are respectively captured, and the accuracy of pet abnormity identification is improved.

Description

Technical Field

[0001] The present application relates to the technical field of pet abnormal behavior recognition, and in particular to a method and system for pet abnormal behavior recognition. Background Art

[0002] Pet locators have been widely used in the daily management of family pets. These devices can help owners track the location of their pets in real time through GPS or other positioning technologies. However, existing pet locators usually only have basic positioning functions and cannot accurately identify whether the current pet behavior is abnormal, resulting in the inability to promptly remind users to check the pet's status and ensure the safety of the pet.

[0003] The Chinese patent "A pet behavior detection system based on image changes", publication number: CN104969875A, publication date: October 14, 2015, specifically discloses: In the first step, the camera module collects image information and transmits it to the intelligent analysis module and the storage module through the transmission module; in the second step, after the intelligent analysis module receives the original signal from the transmission module, it first pre-processes the image, including background subtraction, background extraction, model matching, etc.; in the third step, the intelligent analysis module uses background subtraction to extract the pet in the image, determines whether the pet appears in a suspicious location, and then uses the image background extraction to extract the pet. The technology is used to obtain the detected image after filtering out the animals. The corresponding matching between the continuous images is based on texture, shape, position, etc., and the changes in the position or behavior of the pet within a certain period of time can be obtained; the fourth step, the intelligent analysis module analyzes the changes in the furniture in the suspicious location and obtains the changes in the furniture in the continuous images within a short period of time; the fifth step, after the intelligent analysis module obtains the pet's behavior and the changes in the furniture, it calculates the values ​​of variables X1, X2, and X3 for each feature, thereby obtaining the results of the abnormal behavior of the pet; the sixth step, the intelligent analysis module obtains the results of the abnormal behavior of the pet. This solution judges the abnormal results of the pet based on the changes in the furniture, but it cannot accurately identify the abnormal behavior of the pet in the outdoor environment.

[0004] The Chinese patent "A Pet Behavior Analysis System Based on AI Vision", publication number: CN117894078A, publication date: April 16, 2024, specifically discloses: Through deep learning algorithms, the system can intelligently analyze various behaviors of pets, including daily behaviors such as sleeping, eating, playing, and defecation, and abnormal behaviors such as abnormal barking, urinating and defecating anywhere, destroying furniture, and abnormal movement patterns, to achieve accurate identification and analysis of pet behaviors; at the same time, record the daily behavior status of pets, generate a pet health data set, and compare it with real-time data to determine whether the pet has an abnormal health state. This solution learns all behaviors of pets in a unified manner, but the behavioral habits of pets in different states will definitely be different, and the accuracy is poor. Summary of the invention

[0005] In order to solve the problem in the prior art that all behaviors of pets are used for learning, resulting in the mutual influence of different habits of pets in different states and reduced recognition accuracy, this application provides a method and system for identifying abnormal behaviors of pets, which can respectively capture abnormalities of pets in active states and abnormalities in static states, thereby improving the accuracy of identifying abnormalities of pets.

[0006] In order to achieve the above technical purpose, the present application provides a technical solution, which is a method for identifying abnormal behavior of a pet, comprising the following steps: S1: obtaining historical behavior data of the pet, and preprocessing the historical behavior data of the pet to obtain effective behavior data; S2: Divide the effective behavior data into historical dynamic behavior data and historical static behavior data according to the pet's behavior; S3: Calculate the dynamic behavior abnormality coefficient threshold based on the historical dynamic behavior data, and calculate the static behavior abnormality coefficient threshold based on the historical static behavior data; S4: When the pet's current behavior is a dynamic behavior, use the dynamic behavior abnormality coefficient threshold and the pet's current behavior data to perform abnormal behavior judgment on the pet; when the pet's current behavior is a static behavior, use the static behavior abnormality coefficient threshold and the pet's current behavior data to perform abnormal behavior judgment on the pet; S5: Obtain the number of abnormal behaviors of the pet within a preset period. When the number of abnormal behaviors of the pet reaches the abnormal number threshold, the pet is considered abnormal and a prompt is given.

[0007] The present invention is further configured such that the calculation of the dynamic behavior abnormality coefficient threshold based on the historical dynamic behavior data includes: dividing the historical dynamic behavior data into historical dynamic abnormal data and historical dynamic normal data; calculating a first abnormality coefficient of the historical dynamic abnormal data compared to the historical dynamic normal data; and using the minimum abnormality coefficient as the dynamic behavior abnormality coefficient threshold.

[0008] The present invention is further configured such that the calculation of the first abnormality coefficient of the historical dynamic abnormal data compared to the historical dynamic normal data includes: outputting the matching value of the historical dynamic abnormal data and the historical dynamic normal data based on the matching degree of the pet trajectory and the motion sequence; and calculating the first abnormality coefficient using the historical dynamic abnormal data, the historical dynamic normal data and the matching value.

[0009] The present invention is further configured such that the calculation of the abnormal coefficient of the corresponding pet trajectory and motion timing using the historical dynamic abnormal data, the historical dynamic normal data and the matching value includes: constructing the weight coefficient corresponding to the historical dynamic normal data using the matching value; and obtaining the first abnormal coefficient by combining the standard deviation of the historical dynamic abnormal data and the historical dynamic normal data with the weight coefficient.

[0010] The present invention is further configured such that S4 also includes: judging the current behavior state of the pet based on the pet's movement intensity threshold; if the movement intensity data in the pet's current behavior data is greater than the pet's movement intensity threshold, the pet's current behavior is considered to be a dynamic behavior; if the movement intensity data in the pet's current behavior data is less than or equal to the pet's movement intensity threshold, the pet's current behavior is considered to be a static behavior.

[0011] The present invention is further configured such that the pet abnormal behavior judgment based on the dynamic behavior abnormality coefficient threshold and the pet's current behavior includes: calculating the pet's current dynamic behavior abnormality coefficient based on the pet's current behavior data and historical dynamic normal data; when the pet's current dynamic behavior abnormality coefficient is greater than or equal to the dynamic behavior abnormality coefficient threshold, the pet's dynamic behavior is considered abnormal.

[0012] The present invention is further configured such that S1 also includes: preprocessing the pet's historical behavior data, classifying the pet's historical behavior data generated by the pet's own behavior into valid behavior data, and classifying the pet's historical behavior data generated by the user's behavior into invalid behavior data.

[0013] The present invention is further configured such that S4 also includes: judging the current behavior state of the pet based on the pet's movement intensity and trajectory; when there is a movement trajectory in the pet's current behavior data and the movement intensity data in the pet's current behavior data is greater than the pet's movement intensity threshold, the pet's current behavior is considered to belong to the dynamic behavior corresponding to the pet's own behavior; when there is a movement trajectory in the pet's current behavior data and the movement intensity data in the pet's current behavior data is less than or equal to the pet's movement intensity threshold, the pet's current behavior is considered to belong to the user's behavior.

[0014] The present invention is further configured to include: when the pet's current behavior belongs to the user's behavior, the following is executed: S7: outputting the user's habit data according to the invalid behavior data; when the pet's current behavior data does not match the user's habit data, the pet is considered abnormal and a prompt is given.

[0015] Another technical solution provided by the present application is a pet abnormal behavior identification system, which is used to implement the method as described above, including: a data acquisition module, used to collect pet behavior data; a threshold update module, used to update the abnormality coefficient threshold based on historical dynamic behavior data and historical static behavior data; an abnormality analysis module, used to output the pet abnormality judgment result based on the pet's current behavior data and the abnormality coefficient threshold.

[0016] The beneficial effects of the present application are as follows: by dividing the behavior data of the pet in different states, the abnormal coefficient threshold of the behavior data in different states is obtained, and the current behavior data of the pet is compared with the abnormal coefficient threshold of the corresponding state to determine whether the pet has abnormal behavior, and based on the number of abnormal behaviors of the pet within a preset period, it is further determined whether the abnormal behavior of the pet is an occasional abnormality. The static and dynamic behaviors respectively reflect the performance of the pet in different physiological and psychological states. The dynamic behavior abnormal coefficient threshold and the static behavior abnormal coefficient threshold are used to capture the abnormalities of the pet in the active state and the abnormalities in the static state, thereby improving the accuracy of pet abnormality identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of the abnormal behavior identification method for pets in this application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the present application more clear, the present application is further described in detail below in conjunction with the drawings and examples. It should be understood that the specific implementation method described here is only an optimal embodiment of the present application, which is only used to explain the present application and does not limit the scope of protection of the present application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0019] like Figure 1 As shown, as the first embodiment of the present application, the method for identifying abnormal behavior of a pet includes the following steps: S1: Obtaining historical behavior data of the pet, and preprocessing the historical behavior data of the pet to obtain valid behavior data; S2: Divide the valid behavior data into historical dynamic behavior data and historical static behavior data according to the pet behavior; S3: Calculate the dynamic behavior abnormality coefficient threshold based on the historical dynamic behavior data, and calculate the static behavior abnormality coefficient threshold based on the historical static behavior data; S4: When the pet's current behavior is a dynamic behavior, the abnormal behavior of the pet is judged based on the abnormal coefficient threshold of the dynamic behavior and the current behavior data of the pet; when the pet's current behavior is a static behavior, the abnormal behavior of the pet is judged based on the abnormal coefficient threshold of the static behavior and the current behavior data of the pet; S5: Obtain the number of abnormal behaviors of the pet within a preset period. When the number of abnormal behaviors of the pet reaches an abnormal number threshold, the pet is considered abnormal and a prompt is given.

[0020] In this embodiment, the abnormal coefficient threshold of the behavior data in different states is obtained by dividing the behavior data of the pet in different states. The current behavior data of the pet is compared with the abnormal coefficient threshold of the corresponding state to determine whether the pet has abnormal behavior, and based on the number of abnormal behaviors of the pet within a preset period, it is further determined whether the abnormal behavior of the pet is an occasional abnormality. The static and dynamic behaviors respectively reflect the performance of the pet in different physiological and psychological states. The dynamic behavior abnormal coefficient threshold and the static behavior abnormal coefficient threshold are used to capture the abnormalities of the pet in the active state and the abnormalities in the static state, thereby improving the accuracy of pet abnormality identification.

[0021] Specifically, the pet's historical behavior data at least includes the pet's historical trajectory data, the pet's historical running steps, the pet's historical exercise intensity, the pet's historical rest time, and the pet's historical ambient lighting. Perform data preprocessing on the pet's historical behavior data, filter out erroneous data, and obtain the pet's valid behavior data. Depending on whether the pet's behavior is dynamic or static, the valid behavior data is divided into historical dynamic behavior data and historical static behavior data. Dynamic behavior and static behavior can be divided according to the difference in pet exercise intensity, such as high-intensity exercise belongs to dynamic behavior and low-intensity exercise belongs to static behavior. Dynamic behavior and static behavior can also be divided based on whether the pet is in a resting state, such as when the pet is sleeping as a static behavior, and the rest of the time sequence actions are all dynamic behaviors.

[0022] Calculating the dynamic behavior anomaly coefficient threshold based on historical dynamic behavior data includes: Divide the historical dynamic behavior data into historical dynamic abnormal data and historical dynamic normal data; Calculate the first abnormal coefficient of the historical dynamic abnormal data compared with the historical dynamic normal data; The minimum first abnormal coefficient is used as the dynamic behavior abnormal coefficient threshold.

[0023] Calculating the static behavior abnormality coefficient threshold based on historical static behavior data includes: Divide the historical static behavior data into historical static abnormal data and historical static normal data; Calculate the second anomaly coefficient of the historical static anomaly data compared with the historical dynamic static data; The minimum second abnormal coefficient is used as the dynamic behavior abnormal coefficient threshold.

[0024] The minimum fluctuation coefficient of the pet's behavior data when it has dynamic abnormalities compared to the dynamic normal behavior data is used as the dynamic behavior abnormality coefficient threshold, and the minimum fluctuation coefficient of the pet's behavior data when it has static abnormalities compared to the static normal behavior data is used as the static behavior abnormality coefficient threshold. The fluctuation coefficient, that is, the abnormality coefficient, is used to judge the pet's abnormal behavior instead of relying on fixed pet behavior data as the judgment standard, so as to improve the dynamic adaptability of pet abnormal behavior recognition and ensure the accuracy of pet abnormal behavior recognition.

[0025] In this embodiment, calculating the first abnormality coefficient of the historical dynamic abnormal data compared with the historical dynamic normal data includes: Output the matching values ​​of historical dynamic abnormal data and historical dynamic normal data based on the matching degree of pet trajectory and motion sequence; The first abnormality coefficient is calculated using the historical dynamic abnormal data, the historical dynamic normal data and the matching value.

[0026] The matching value of historical dynamic abnormal data and historical dynamic normal data can be obtained by the matching degree of the pet trajectory and the motion sequence. For example, when the sequence is the same but the trajectory is deviated or the trajectory is the same but the sequence is deviated, the matching value of the historical dynamic abnormal data and the historical dynamic normal data is obtained according to the deviation. At this time, the first abnormal coefficient corresponding to the pet trajectory and the motion sequence is calculated by the historical dynamic abnormal data, the historical dynamic normal data and the matching value, including: The weight coefficient corresponding to the historical dynamic normal data is constructed with the matching value; The first abnormal coefficient is obtained by combining the standard deviation of the historical dynamic abnormal data and the historical dynamic normal data with the weight coefficient.

[0027] Taking a certain historical dynamic anomaly as an example, according to the time node and trajectory of the dynamic anomaly, the trajectory shows the pet's position, environment and movement bias at that time, and matches it with the time node and trajectory of all historical dynamic normal data, calculates the matching value, and uses the matching value to construct the weight coefficient corresponding to the historical dynamic normal data, that is, the more consistent the time node and trajectory are, the greater the weight coefficient of the historical dynamic normal data is. At this time, the value calculated by the standard deviation of the historical dynamic normal data and the historical dynamic abnormal data should account for a larger proportion in the abnormal coefficient, because it reflects a more consistent difference between abnormal behavior and normal behavior. All historical dynamic normal data can show the global abnormal impact by influencing the abnormal coefficient according to the matching degree. For example, there must be factors with greater influence in the time series and trajectory. Recording the impact with the historical dynamic normal data with deviations can not only ensure that the accuracy of the first abnormal coefficient will not be affected by the existence of the weight coefficient, but also ensure that the global abnormal situation is reflected in the first abnormal coefficient.

[0028] Similarly, calculating the second abnormality coefficient of the historical static abnormal data compared with the historical static normal data includes: Output the matching values ​​of historical static abnormal data and historical static normal data based on the matching degree of pet trajectory and motion sequence; The second abnormality coefficient is calculated using the historical static abnormal data, the historical static normal data and the matching value.

[0029] The second abnormality coefficient corresponding to the pet trajectory and motion sequence is calculated using the historical static abnormal data, the historical static normal data and the matching value, including: The weight coefficient corresponding to the historical static normal data is constructed with the matching value; The second abnormality coefficient is obtained by combining the standard deviation of the historical static abnormal data and the historical static normal data with a weight coefficient.

[0030] In this embodiment, S4 also includes: Determine the pet's current behavior state based on the pet's movement intensity threshold; If the motion intensity data in the pet's current behavior data is greater than the pet's motion intensity threshold, the pet's current behavior is considered to be a dynamic behavior; If the motion intensity data in the pet's current behavior data is less than or equal to the pet's motion intensity threshold, the pet's current behavior is considered to be a static behavior.

[0031] The pet exercise intensity threshold can be set according to the maximum value of static behavior exercise intensity in the pet's historical behavior data, so as to adapt the pet exercise intensity threshold based on individual differences of the pet. At the same time, the pet's exercise ability may change with time, health status and other factors. Therefore, the pet exercise intensity threshold can be dynamically adjusted according to the actual situation to ensure the timeliness of the judgment result, thereby improving the accuracy of the behavior status judgment.

[0032] The abnormal behavior judgment of the pet based on the dynamic abnormal behavior coefficient threshold and the current behavior of the pet includes: Calculate the pet's current dynamic behavior abnormality coefficient based on the pet's current behavior data and historical dynamic normal data; When the pet's current dynamic behavior abnormality coefficient is greater than or equal to the dynamic behavior abnormality coefficient threshold, the pet's dynamic behavior is considered abnormal.

[0033] Calculating the abnormal coefficient of the pet's current dynamic behavior based on the pet's current behavior data and historical dynamic normal data includes: outputting a matching value of the pet's current behavior data and historical dynamic normal data based on the matching degree of the pet's trajectory and the motion sequence; The pet's current dynamic behavior abnormality coefficient is calculated based on the pet's current behavior data, historical dynamic normal data and matching values.

[0034] Based on local matching and global matching, the abnormalities of the pet's current dynamic behavior compared with the same time sequence and the same trajectory, as well as the abnormalities of the pet's current dynamic behavior compared with all time sequences and all trajectories are obtained, so as to avoid misjudgment caused by occasional abnormalities of the pet's current dynamic behavior in the same time sequence and the same trajectory, and improve the accuracy of pet abnormality identification.

[0035] Similarly, performing pet abnormal behavior judgment based on the static behavior abnormality coefficient threshold and the pet's current behavior includes: calculating the pet's current static behavior abnormality coefficient based on the pet's current behavior data and historical static normal data; When the pet's current static behavior abnormality coefficient is greater than or equal to the static behavior abnormality coefficient threshold, the pet's static behavior is considered abnormal.

[0036] Calculating the pet's current static behavior abnormality coefficient based on the pet's current behavior data and historical static normal data includes: outputting a matching value of the pet's current behavior data and historical static normal data based on the matching degree of the pet's trajectory and motion sequence; The pet's current static behavior abnormality coefficient is calculated based on the pet's current behavior data, historical static normal data and matching values.

[0037] Furthermore, when the number of abnormal behaviors of the pet reaches the abnormal number threshold within the preset period, the pet is considered abnormal and the user is prompted to check the pet status. The preset period and abnormal number threshold can be set according to the type of pet, and each prompt situation can also be recorded in the initial stage, and adaptive changes are made according to the prompt situation and whether there is indeed an abnormality. For example, in the initial stage, the preset period is set to 10 minutes and the abnormal number threshold is 2 times. In the data recorded in the subsequent process, it is found that there are no actual abnormalities in 2 abnormalities within 10 minutes, but there are actual abnormalities when the number of abnormalities reaches 3 times within half an hour. According to the recorded data, the preset period is changed to 30 minutes and the abnormal number threshold is 3 times. It can be understood that in the initial situation, the pet historical behavior data can be based on the pet historical behavior data of the corresponding pet type, and in the user's use process, the dynamic behavior abnormality coefficient threshold and the static behavior abnormality coefficient threshold are updated based on the historical behavior data of the user's pet.

[0038] As a second embodiment of the present application, step S1 further includes: The pet historical behavior data is preprocessed, and the pet historical behavior data generated by the pet's own behavior is classified as valid behavior data, and the pet historical behavior data generated by the user's behavior is classified as invalid behavior data.

[0039] In this embodiment, considering that the user may have an impact on the pet's behavior, such as the behavior data when the user holds the pet while walking is bound to be quite different from the behavior data of the pet's own walking, the pet's historical behavior data caused by the user's behavior is filtered out to improve the accuracy of judgment.

[0040] In this case, step S4 further includes: Determine the pet's current behavior status based on the pet's movement intensity and trajectory; When there is a motion track in the pet's current behavior data and the motion intensity data in the pet's current behavior data is greater than the pet's motion intensity threshold, the pet's current behavior is considered to be a dynamic behavior corresponding to the pet's own behavior; When there is a motion track in the pet's current behavior data and the motion intensity data in the pet's current behavior data is less than or equal to the pet's motion intensity threshold, the pet's current behavior is considered to be user behavior.

[0041] In this embodiment, the method for identifying abnormal pet behavior also includes: when the pet's current behavior belongs to user behavior, executing: S7: outputting user habit data based on invalid behavior data, and when the pet's current behavior data does not match the user habit data, the pet is considered abnormal and a prompt is given.

[0042] User habit data at least includes movement trajectory and movement speed. When the pet's current behavior data is affected by human factors, it is determined whether the human impact is consistent with the user's own habits. If not, it is considered that the human impact is not an execution prompt made by the user.

[0043] As a third embodiment of the present application, a pet abnormal behavior recognition system includes: Data collection module, used to collect pet behavior data; A threshold updating module, used to update the abnormal coefficient threshold based on historical dynamic behavior data and historical static behavior data; The abnormality analysis module is used to output the pet abnormality judgment result based on the pet's current behavior data and the abnormality coefficient threshold.

[0044] In this embodiment, the data acquisition module is a locator, which has built-in high-precision sensors such as a three-axis accelerometer, a gyroscope, and an ambient light sensor for real-time collection of the pet's motion data and environmental information. The motion data includes the pet's steps, exercise intensity, rest time, ambient light, etc., so as to comprehensively analyze the pet's behavior and health status. The locator is designed with an anti-disassembly structure and can only be removed by special tools or authorized operations of the application to prevent human disassembly or damage. The locator has IP67-level waterproof and dustproof capabilities and is suitable for a variety of outdoor environments, ensuring that the device can still work stably under harsh conditions such as rainy days and muddy conditions.

[0045] The pet abnormal behavior recognition system also includes: The user interaction module is used to interact with the user based on the pet abnormality judgment result.

[0046] The user interface displays key information such as the pet's real-time location, behavior status, remaining battery, etc. It provides query and review functions for pet behavior history and location information, and supports exporting data for further analysis. Users can customize settings such as alarm sound effects, vibration mode, notification frequency, etc. through the application to meet personal preferences. The application supports multiple user accounts to bind to the same device, allowing family members to share pet management information.

[0047] In other embodiments, the pet abnormal behavior recognition system further includes: Fence building module for drawing geofences.

[0048] At this time, the abnormal analysis module determines whether the pet exceeds the preset range based on the pet behavior data and the geofence output by the fence construction module. Users draw geofences on the map through the supporting mobile phone application to define the area where the pet is allowed to move. It supports circular, polygonal or freely drawn fences. It allows multiple geofences to be set and can be temporarily or permanently adjusted as needed, providing fence management in multiple scenarios such as home and pet park.

[0049] At this time, the user interaction module includes: an alarm unit for executing prompts. The alarm unit can be a buzzer and an LED flash. Once the pet exceeds the preset fence range, the locator will activate the built-in buzzer and LED flash to issue an audible and visual alarm and push an out-of-bounds notification to the owner's mobile phone through the application, including the pet's current GPS location, timestamp and historical trajectory. Users can set the alarm level of different fences (for example, low level near home, high level in unfamiliar areas) to meet the safety needs in different scenarios.

[0050] The specific implementation described above is a preferred implementation of the pet abnormal behavior identification method and system of the present application, and is not intended to limit the specific implementation scope of the present application. The scope of the present application includes but is not limited to the specific implementation embodiment. All equivalent changes made in accordance with the shape and structure of the present application are within the protection scope of the present application.

Claims

1. A method for identifying abnormal behavior of pets, characterized by: The steps include: S1: Obtaining historical behavior data of the pet, and preprocessing the historical behavior data of the pet to obtain valid behavior data; S2: Divide the valid behavior data into historical dynamic behavior data and historical static behavior data according to the pet behavior; S3: Calculate the dynamic behavior abnormality coefficient threshold based on the historical dynamic behavior data, and calculate the static behavior abnormality coefficient threshold based on the historical static behavior data; S4: When the pet's current behavior is a dynamic behavior, the abnormal behavior of the pet is judged based on the abnormal coefficient threshold of the dynamic behavior and the current behavior data of the pet; When the pet's current behavior is a static behavior, the pet's abnormal behavior is judged based on the static behavior abnormality coefficient threshold and the pet's current behavior data; S5: Obtain the number of abnormal behaviors of the pet within a preset period. When the number of abnormal behaviors of the pet reaches an abnormal number threshold, the pet is considered abnormal and a prompt is given.

2. The method for identifying abnormal behavior of a pet according to claim 1, characterized in that: The calculation of the dynamic behavior abnormality coefficient threshold based on the historical dynamic behavior data includes: Divide the historical dynamic behavior data into historical dynamic abnormal data and historical dynamic normal data; Calculate the first abnormal coefficient of the historical dynamic abnormal data compared with the historical dynamic normal data; The minimum abnormal coefficient is used as the dynamic behavior abnormal coefficient threshold.

3. The method for identifying abnormal behavior of a pet according to claim 2, characterized in that: The calculating of the first abnormality coefficient of the historical dynamic abnormal data compared with the historical dynamic normal data comprises: Output the matching values ​​of historical dynamic abnormal data and historical dynamic normal data based on the matching degree of pet trajectory and motion sequence; The first abnormality coefficient is calculated using the historical dynamic abnormal data, the historical dynamic normal data and the matching value.

4. The method for identifying abnormal behavior of a pet according to claim 3, characterized in that: The calculation of the abnormal coefficient of the corresponding pet trajectory and motion sequence using the historical dynamic abnormal data, the historical dynamic normal data and the matching value includes: The weight coefficient corresponding to the historical dynamic normal data is constructed with the matching value; The first abnormal coefficient is obtained by combining the standard deviation of the historical dynamic abnormal data and the historical dynamic normal data with the weight coefficient.

5. The method for identifying abnormal behavior of a pet according to claim 1, characterized in that: The S4 further comprises: Determine the pet's current behavior state based on the pet's movement intensity threshold; If the motion intensity data in the pet's current behavior data is greater than the pet's motion intensity threshold, the pet's current behavior is considered to be a dynamic behavior; If the motion intensity data in the pet's current behavior data is less than or equal to the pet's motion intensity threshold, the pet's current behavior is considered to be a static behavior.

6. The method for identifying abnormal behavior of a pet according to claim 2, characterized in that: The determination of abnormal behavior of the pet based on the dynamic abnormal behavior coefficient threshold and the current behavior of the pet includes: Calculate the pet's current dynamic behavior abnormality coefficient based on the pet's current behavior data and historical dynamic normal data; When the pet's current dynamic behavior abnormality coefficient is greater than or equal to the dynamic behavior abnormality coefficient threshold, the pet's dynamic behavior is considered abnormal.

7. The method for identifying abnormal behavior of a pet according to claim 1, characterized in that: The S1 further comprises: The pet historical behavior data is preprocessed, and the pet historical behavior data generated by the pet's own behavior is classified as valid behavior data, and the pet historical behavior data generated by the user's behavior is classified as invalid behavior data.

8. The method for identifying abnormal behavior of a pet according to claim 7, characterized in that: The S4 further comprises: Determine the pet's current behavior status based on the pet's movement intensity and trajectory; When there is a motion track in the pet's current behavior data and the motion intensity data in the pet's current behavior data is greater than the pet's motion intensity threshold, the pet's current behavior is considered to be a dynamic behavior corresponding to the pet's own behavior; When there is a motion track in the pet's current behavior data and the motion intensity data in the pet's current behavior data is less than or equal to the pet's motion intensity threshold, the pet's current behavior is considered to be user behavior.

9. The method for identifying abnormal behavior of a pet according to claim 8, characterized in that: Also includes: When the pet's current behavior is a user behavior execution: S7: Output the user habit data according to the invalid behavior data. When the pet's current behavior data does not match the user's habit data, the pet is considered abnormal and a prompt is given.

10. A pet abnormal behavior recognition system, used to implement the pet abnormal behavior recognition method according to any one of claims 1 to 9, characterized in that: The pet abnormal behavior recognition system includes: Data collection module, used to collect pet behavior data; A threshold updating module, used to update the abnormal coefficient threshold based on historical dynamic behavior data and historical static behavior data; The abnormality analysis module is used to output the pet abnormality judgment result based on the pet's current behavior data and the abnormality coefficient threshold.

Images