Pet health monitoring system and method based on computer vision and biosensing

Abstract

The invention provides a pet health monitoring system based on computer vision and biosensing. The pet health monitoring system comprises an input layer, a processing layer and an output layer, the input layer is a multi-modal data acquisition module, and the multi-modal data acquisition module comprises a biosensor array unit and a visual acquisition unit; the visual acquisition unit is used for capturing visual information of behaviors and appearances of pets, converting the visual information into digital signals and outputting the digital signals; the processing layer comprises an edge calculation module, and the edge calculation module is used for data transmitted by the biosensor array unit and the visual acquisition unit. According to the pet health monitoring system based on computer vision and biosensing, the multi-modal data acquisition module, the edge calculation module and the intelligent interaction module are utilized, compared with traditional manual observation, the abnormal behavior detection accuracy reaches 92%, early disease early warning can be achieved 24-48 hours in advance, and the pet health monitoring system based on computer vision and biosensing has the advantages of being high in practicability and the like. And precious time is won for timely treatment of pets.

Description

Technical Field

[0001] The present invention relates to the field of intelligent pet devices, and particularly to a pet health monitoring system and method based on computer vision and biosensing. Background Art

[0002] In the field of pet health monitoring, with the continuous growth of the number of pet breedings and the increasing attention of pet owners to pet health, the development of pet health monitoring technology has become increasingly important.

[0003] However, there are many problems to be solved urgently in the current related technologies:

[0004] On the one hand, the functions of existing pet health monitoring devices are relatively single. For example, the common independent body temperature monitoring collars on the market only have the function of body temperature monitoring and cannot obtain other aspects of pet health information; ordinary cameras can only be used to observe the appearance of pets and are difficult to deeply analyze the health status of pets. Due to the lack of the ability to synergistically analyze multi-dimensional data, it is impossible to comprehensively and accurately evaluate the health status of pets and it is difficult to meet the needs of pet owners for refined management of pet health.

[0005] On the other hand, traditional pet guardianship methods mainly rely on manual observation. This method has obvious limitations. It is not only inefficient, but also difficult to capture abnormal behaviors of pets in real time, such as limping and anorexia. At the same time, it is also impossible to monitor changes in pet physiological indicators in a timely manner, such as abnormal heart rate, etc., resulting in potential pet health problems not being discovered in time and delaying the best treatment opportunity.

[0006] In addition, existing pet health monitoring devices also have deficiencies in battery life and data privacy protection. Most pet health monitoring collars have a short battery life, and frequent charging brings great inconvenience to users, reducing the continuity of device use and user experience. At the same time, the data privacy protection mechanism is imperfect, and there is a risk of leakage of pet health data during the process of data collection, transmission, and storage, which may have a negative impact on the privacy of pet owners and the health management of pets.

[0007] Therefore, it is necessary to provide a pet health monitoring system based on computer vision and biosensing to solve the above technical problems. Summary of the Invention

[0008] The present invention provides a pet health monitoring system based on computer vision and biosensing, which solves the problems that the monitoring devices of the current device have a single function, lack the ability to synergistically analyze multi-dimensional data, cannot comprehensively evaluate the health status of pets, the traditional manual guardianship method is inefficient, it is difficult to capture abnormal behaviors and physiological index changes of pets in real time, it is easy to delay the discovery of health problems, the monitoring device has insufficient battery life, frequent charging affects the use experience, and the data privacy protection mechanism is imperfect and there is a risk of data leakage.

[0009] To solve the above technical problems, the pet health monitoring system based on computer vision and biosensing provided by the present invention includes: an input layer, a processing layer, and an output layer;

[0010] The input layer is a multimodal data acquisition module, and the multimodal data acquisition module includes a biosensor array unit and a vision acquisition unit;

[0011] The vision acquisition unit is used to capture visual information of the pet's behavior and appearance, and convert it into a digital signal for output;

[0012] The processing layer includes an edge computing module, and the edge computing module is used for the data transmitted by the biosensor array unit and the vision acquisition unit;

[0013] The output layer is used to receive the results of the processing layer and perform feedback actions, including an intelligent interaction module, and the intelligent interaction module is used to receive the judgment results of the edge computing unit.

[0014] Preferably, the edge computing unit includes an integrated spatio-temporal feature fusion model and an anomaly detection algorithm. The integrated spatio-temporal feature fusion model is used to fuse the spatio-temporal features of data from different sources, and the anomaly detection algorithm is used to analyze the processed data to determine whether the pet's health status is abnormal.

[0015] Preferably, the biosensor array unit includes a graphene-based flexible body temperature sensor and a dry electrode type EMG sensor. The graphene-based flexible body temperature sensor is used to accurately measure the pet's body temperature with an accuracy of ±0.1°C and a response time < 2s, and transmit the body temperature data. The dry electrode type EMG sensor is used to collect the chewing myoelectric signals of the pet to judge the pet's feeding state and output relevant data.

[0016] Preferably, the graphene-based flexible body temperature sensor is made of graphene material to achieve high-precision and fast-response body temperature measurement.

[0017] Preferably, the dry electrode type EMG sensor can directly collect the chewing myoelectric signals of the pet without using auxiliary materials such as conductive paste.

[0018] Preferably, the vision acquisition unit includes a high-definition camera and an image processor, which are used to capture clear pet visual information and perform preliminary processing.

[0019] Preferably, the edge computer module includes a vibration feedback unit and an APP warning push unit. The vibration feedback unit uses a micro vibration motor and can generate vibration signals of different intensities and frequencies. The APP warning push unit is used to push pet health warning information to the user's mobile APP in real time through network communication technology.

[0020] A pet health monitoring device based on computer vision and biosensing, comprising: an instrument body, a display screen, a camera, a switch, a charging port, two connecting rings, a strap, and two magic tapes. The display screen is arranged on the surface of the instrument body, the camera is arranged on the surface of the instrument body, the switch and the charging port are respectively installed on one side of the surface of the instrument body, the two connecting rings are symmetrically installed on both sides of the back of the instrument body, the strap is arranged between the two connecting rings, and the two magic tapes are arranged on both sides of the surface of the strap in a staggered manner.

[0021] An installation component is arranged between the instrument body and the connecting ring. The installation component includes an installation sleeve, an installation block, and a fixing bolt. The installation sleeve is connected to the back of the instrument body, the installation block is connected to the inside of the installation sleeve in a plug-and-play manner, and the fixing bolt is arranged between the installation sleeve and the installation block.

[0022] A pet health monitoring method based on computer vision and biosensing includes the following steps:

[0023] S1. Data acquisition stage;

[0024] S11. Biosensing data acquisition: The flexible temperature sensor based on graphene exerts the characteristics of high precision and fast response, accurately measures the pet's body temperature under the conditions of an accuracy of ±0.1°C and a response time <2 s, and transmits the body temperature data. The dry electrode type EMG sensor directly collects the electromyogram signal of the pet's chewing muscle to judge the pet's eating state and output relevant data;

[0025] S12. Visual data acquisition: The high-definition camera in the visual acquisition unit captures the visual information of the pet's behavior and appearance, and the image processor converts this visual information into digital signals and outputs them;

[0026] S2. Data processing stage;

[0027] S21. Data aggregation: The body temperature, eating state and other data collected by the biosensor array unit and the visual digital signals output by the visual acquisition unit are all transmitted to the edge computing module in the processing layer;

[0028] S22. Feature fusion: The integrated spatio-temporal feature fusion model in the edge computing module performs spatio-temporal feature fusion on data from different sources to integrate multi-dimensional information;

[0029] S23. Health judgment: The anomaly detection algorithm analyzes the data after fusion processing to determine whether the pet's health condition is abnormal;

[0030] S3. Result feedback stage;

[0031] S31. Result reception: If the edge computing module determines that the pet's health condition is abnormal, the intelligent interaction module in the output layer receives this judgment result;

[0032] S32. Vibration feedback: The vibration feedback unit in the intelligent interaction module generates vibration signals with different intensities and frequencies to remind relevant personnel;

[0033] S33. Early warning push: The APP early warning push unit uses network communication technology to push pet health early warning information to the user's mobile APP in real time, enabling the user to timely understand the pet's health condition.

[0034] Compared with related technologies, the pet health monitoring system based on computer vision and biosensing provided by the present invention has the following beneficial effects:

[0035] The present invention provides a pet health monitoring system based on computer vision and biosensing, which uses a multi-modal data acquisition module, an edge computing module, and an intelligent interaction module. Compared with traditional manual observation, the anomaly behavior detection accuracy rate of the present invention reaches 92%, and it can realize early disease warning 24 - 48 hours in advance, striving for precious time for the timely treatment of pets. In terms of data privacy protection, the federated learning technology is adopted, and sensitive physiological data only completes feature extraction at the device end, and the original data is not uploaded to the cloud, effectively ensuring the security of pet health data. In terms of wearing comfort, the weight of the collar main body is <50g, and the sensor contact surface uses medical-grade silica gel to ensure that the pet wears comfortably and safely. Brief Description of the Drawings

[0036] Figure 1 It is a schematic structural diagram of the first embodiment of the pet health monitoring system based on computer vision and biosensing provided by the present invention;

[0037] Figure 2 It is a schematic structural diagram of the second embodiment of the pet health monitoring system based on computer vision and biosensing provided by the present invention;

[0038] Figure 3 is Figure 2 a three-dimensional structural diagram of the device shown;

[0039] Figure 4 is Figure 2 a rear view of the device shown;

[0040] Figure 5 is Figure 4Schematic enlarged view of part A shown

[0041] Figure 6 Schematic structural diagram of the third embodiment of the pet health monitoring system based on computer vision and biosensing provided by the present invention

[0042] Reference numerals in the figure: 1, instrument main body; 2, display screen; 3, camera; 4, switch; 5, charging port; 6, connecting ring; 7, strap; 8, magic tape; 9, installation component; 91, installation sleeve; 92, installation block; 93, fixing bolt; 10, protection component; 101, collar; 102, sponge pad; 103, rubber pad Specific implementation manner

[0043] The present invention will be further described below with reference to the drawings and embodiments

[0044] First embodiment

[0045] Please refer to Figure 1 , wherein Figure 1 Schematic structural diagram of the first embodiment of the pet health monitoring system based on computer vision and biosensing provided by the present invention. The pet health monitoring system based on computer vision and biosensing includes: an input layer, a processing layer, and an output layer

[0046] The input layer is a multimodal data acquisition module, and the multimodal data acquisition module includes a biosensor array unit and a vision acquisition unit

[0047] The vision acquisition unit is used to capture visual information of pet behavior and appearance and convert it into digital signals for output

[0048] The processing layer includes an edge computing module, and the edge computing module is used for the data transmitted by the biosensor array unit and the vision acquisition unit

[0049] The output layer is used to receive the results of the processing layer and perform feedback actions, including an intelligent interaction module, and the intelligent interaction module is used to receive the judgment results of the edge computing unit

[0050] The edge computing unit includes an integrated spatio-temporal feature fusion model and an anomaly detection algorithm. The integrated spatio-temporal feature fusion model is used to fuse the spatio-temporal features of data from different sources, and the anomaly detection algorithm is used to analyze the processed data to judge whether the pet health status is abnormal

[0051] The biosensor array unit includes a graphene-based flexible body temperature sensor and a dry electrode EMG sensor. The graphene-based flexible body temperature sensor is used to accurately measure the pet's body temperature with an accuracy of ±0.1°C and a response time of <2s, and transmit the body temperature data. The dry electrode EMG sensor is used to collect the pet's chewing electromyographic signal to determine the pet's eating status and output relevant data.

[0052] The graphene-based flexible body temperature sensor is made of graphene material to achieve high-precision and fast-response body temperature measurement.

[0053] The dry electrode EMG sensor can directly collect the electromyographic signal of pet chewing without using auxiliary materials such as conductive paste.

[0054] The visual acquisition unit includes a high-definition camera and an image processor, which are used to capture clear pet visual information and perform preliminary processing.

[0055] The edge computer module includes a vibration feedback unit and an APP warning push unit. The vibration feedback unit adopts a micro vibration motor and can generate vibration signals of different intensities and frequencies. The APP warning push unit is used for the warning push module to push the pet health warning information to the user's mobile phone APP in real time through network communication technology.

[0056] The graphene flexible body temperature sensor in the biosensor array unit uses graphene's good thermal and electrical conductivity to quickly and accurately measure the pet's body temperature, with a short response time, and can promptly reflect changes in body temperature. The dry electrode EMG sensor directly collects chewing electromyographic signals without the need for conductive paste, avoiding the inconvenience and hygiene problems of traditional wet electrodes, and accurately judging the pet's eating status.

[0057] The high-definition camera of the visual acquisition unit captures pet behavior and appearance details from multiple angles, and the image processor quickly processes the images and converts them into digital signals, providing rich information for subsequent analysis.

[0058] After receiving multi-source data, the edge computing module integrates the spatiotemporal feature fusion model to deeply fuse the data. For example, it associates the changes in the pet's body temperature during eating with the visual data of the eating action to mine hidden information. The anomaly detection algorithm analyzes the fused data and combines it with the preset health model to judge the pet's health status. If the body temperature persists to be abnormal and is accompanied by behaviors such as loss of appetite, an abnormal signal will be issued in time.

[0059] After the intelligent interaction module receives an abnormal signal, the vibration feedback unit adjusts the vibration intensity and frequency according to the degree of abnormality to give different degrees of reminders to the pet owner. The APP early warning push unit uses network communication technology to push detailed abnormal information, such as body temperature values, abnormal behavior descriptions, etc., to the user's mobile APP, facilitating the owner to take measures in a timely manner.

[0060] Compared with related technologies, the pet health monitoring system based on computer vision and biosensing provided by the present invention has the following beneficial effects:

[0061] The present invention provides a pet health monitoring system based on computer vision and biosensing, which uses a multi-modal data acquisition module, an edge computing module, and an intelligent interaction module. Compared with traditional manual observation, the accuracy rate of abnormal behavior detection of the present invention reaches 92%, and early disease warning can be realized 24-48 hours in advance, striving for precious time for the timely treatment of pets. In terms of data privacy protection, federated learning technology is adopted, and sensitive physiological data only completes feature extraction at the device end, and the original data is not uploaded to the cloud, effectively ensuring the security of pet health data. In terms of wearing comfort, the weight of the collar main body is <50g, and the sensor contact surface uses medical-grade silicone to ensure that the pet wears comfortably and safely.

[0062] Second Embodiment

[0063] Please refer to Figure 2 、 Figure 3 、 Figure 4 and Figure 5 Based on the pet health monitoring system based on computer vision and biosensing provided in the first embodiment of the present application, the second embodiment of the present application proposes another pet health monitoring system based on computer vision and biosensing. The second embodiment is only a preferred manner of the first embodiment, and the implementation of the second embodiment will not affect the independent implementation of the first embodiment.

[0064] Specifically, the difference between the pet health monitoring system based on computer vision and biosensing provided in the second embodiment of the present application is that the pet health monitoring system based on computer vision and biosensing includes: an instrument main body 1, a display screen 2, a camera 3, a switch 4, a charging port 5, two connecting rings 6, a strap 7, and two magic tapes 8. The display screen 2 is arranged on the surface of the instrument main body 1, the camera 3 is arranged on the surface of the instrument main body 1, the switch 4 and the charging port 5 are respectively installed on one side of the surface of the instrument main body 1, the two connecting rings 6 are symmetrically installed on both sides of the back of the instrument main body 1, the strap 7 is arranged between the two connecting rings 6, and the two magic tapes 8 are arranged on both sides of the surface of the strap 7 in a staggered manner.

[0065] An installation component 9 is provided between the instrument main body 1 and the connection ring 6. The installation component 9 includes an installation sleeve 91, an installation block 92, and a fixing bolt 93. The installation sleeve 91 is connected to the back of the instrument main body 1. The installation block 92 is inserted and removed in the installation sleeve 91. The fixing bolt 93 is arranged between the installation sleeve 91 and the installation block 92.

[0066] The display screen 2 on the instrument main body can display some pet health data in real time, such as body temperature, eating status icon, etc., which is convenient for the owner to check at any time. The camera 3 is used to collect visual data. The switch control 4 starts and stops the device. The charging port 5 replenishes the power of the device.

[0067] The two connection rings 6 and the strap 7 cooperate. The length of the strap is adjusted through the magic tape 8 to fix the instrument main body on the pet. The installation component makes the installation and disassembly of the connection ring 6 and the instrument main body 1 more convenient. When it is necessary to change the position of the device or perform maintenance, the fixing bolt 93 can be unscrewed, the installation block 92 can be pulled out, and the connection ring 6 and the instrument main body 1 can be easily separated.

[0068] In actual use, a miniaturized power management system is integrated inside the instrument main body, which can optimize the battery charging and discharging process and extend the battery life of the device. The display screen uses a low-power and high-contrast screen material, which reduces energy consumption while ensuring clear display. The camera has an autofocus and light adaptability function, and can obtain clear images under different light conditions. In addition, to improve the stability and reliability of the device, the circuit design inside the instrument main body uses a multi-layer circuit board to reduce signal interference and ensure the accuracy of data transmission.

[0069] Hardware composition

[0070] The collar main body integrates a three-axis magnetometer, a flexible ECG electrode, and a 940nm infrared camera (resolution 640×480), and can collect the pet's motion state, ECG signal, and behavior images.

[0071] The environmental base station is equipped with an odor sensor (detecting ammonia concentration) and a laser ranging module (monitoring the pet's activity range), which are used to monitor the pet's living environment and activity conditions.

[0072] Workflow

[0073] Every hour, a 5-minute eating video is collected through the infrared camera, the remaining amount in the food bowl is identified using OCR technology, and the feeding plan is dynamically adjusted in combination with the weight data to achieve scientific feeding.

[0074] When the odor sensor detects an abnormal odor of excrement (> threshold 50ppm), the camera is triggered to take an image of the excrement, and the color / morphology abnormality is analyzed through the YOLOv5 model to timely detect the pet's excretion problem.

[0075] Compared with the related technologies, the pet health monitoring system based on computer vision and biosensing provided by the present invention has the following beneficial effects:

[0076] The present invention provides a pet health monitoring system based on computer vision and biosensing. A strap 7 with two Velcro fasteners 8 is provided on an instrument body 1 with a display screen 2, a camera 3, a switch 4, a charging port 5, and two connecting rings 6, which facilitates the use of the instrument body 1 with the system on a pet. An installation sleeve 91, an installation block 92, and a fixing bolt 93 are provided between the instrument body 1 and the connecting ring 6 to facilitate the convenient installation and disassembly between the connecting ring 6 and the instrument body 1.

[0077] Third Embodiment

[0078] Please refer to Figure 6 , based on the pet health monitoring system based on computer vision and biosensing provided in the first embodiment of the present application, the third embodiment of the present application proposes another pet health monitoring system based on computer vision and biosensing. The third embodiment is only a preferred manner of the first embodiment, and the implementation of the third embodiment will not affect the independent implementation of the first embodiment.

[0079] Specifically, the difference of the pet health monitoring system based on computer vision and biosensing provided in the third embodiment of the present application is that, for the pet health monitoring system based on computer vision and biosensing, a protection component 10 is provided on the surface of the strap 7, and the protection component 10 includes two sleeve rings 101, a sponge pad 102, and two rubber pads 103. The two rubber pads 103 are respectively bonded inside the two sleeve rings 101, the two sleeve rings 101 are sleeved on the surface of the strap 7, and the sponge pad 102 is bonded between the two sleeve rings 101.

[0080] When the protection component is provided on the surface of the strap, when the device is worn around the pet's neck, the sleeve rings and the rubber pads fit the pet's neck skin, dispersing the pressure of the strap, and the sponge pad plays a buffering role to avoid friction or marks on the pet's neck caused by the strap, improving the comfort of the pet wearing.

[0081] The sleeve rings 101 and the rubber pads 103 of the protection component 10 are made of a soft, skin-friendly, and breathable environmental protection material, such as medical-grade silicone, which not only does not irritate the pet's skin but also prevents skin problems caused by sweat accumulation. The sponge pad 102 is made of a high-elasticity, slow-rebound sponge and is not easily deformed after long-term use. Some micro-protrusions or textures can be designed on the surfaces of the sleeve rings 101 and the rubber pads 103 to increase the friction with the pet's neck and prevent the device from shaking or shifting when the pet moves. In addition, the protection component can be designed into a detachable structure for easy cleaning and replacement to maintain the hygiene and performance of the device.

[0082] Compared with the related technologies, the pet health monitoring system based on computer vision and biosensing provided by the present invention has the following beneficial effects:

[0083] The present invention provides a pet health monitoring system based on computer vision and biosensing. A protection component 10 is arranged on the surface of the strap 7, which can increase the comfort when contacting and fixing the pet's neck.

[0084] A pet health monitoring method based on computer vision and biosensing includes the following steps:

[0085] S1. Data acquisition stage;

[0086] S11. Biosensing data acquisition: The flexible temperature sensor based on graphene exerts the characteristics of high precision and fast response, accurately measures the pet's body temperature under the conditions of an accuracy of ±0.1°C and a response time <2 s, and transmits the body temperature data. The dry electrode type EMG sensor directly collects the pet's masticatory myoelectric signal to judge the pet's eating state and output relevant data;

[0087] S12. Visual data acquisition: The high-definition camera in the visual acquisition unit captures the visual information of the pet's behavior and appearance, and the image processor converts this visual information into digital signals for output;

[0088] S2. Data processing stage;

[0089] S21. Data aggregation: The body temperature, eating state and other data collected by the biosensor array unit and the visual digital signals output by the visual acquisition unit are all transmitted to the edge computing module in the processing layer;

[0090] S22. Feature fusion: The integrated spatio-temporal feature fusion model in the edge computing module performs spatio-temporal feature fusion on data from different sources to integrate multi-dimensional information;

[0091] S23. Health judgment: The anomaly detection algorithm analyzes the data after fusion processing to judge whether the pet's health condition is abnormal;

[0092] S3. Result feedback stage;

[0093] S31. Result reception: If the edge computing module judges that the pet's health condition is abnormal, the intelligent interaction module in the output layer receives this judgment result;

[0094] S32. Vibration feedback: The vibration feedback unit in the intelligent interaction module generates vibration signals of different intensities and frequencies to remind relevant personnel;

[0095] S33. Early warning push: The APP early warning push unit uses network communication technology to push the pet health early warning information to the user's mobile APP in real time, so that the user can timely understand the pet's health condition.

[0096] The above are only embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be similarly included in the patent protection scope of the present invention.

Claims

1. A pet health monitoring system based on computer vision and biosensing, characterized in that, Comprising: An input layer, a processing layer, and an output layer; The input layer is a multi-modal data acquisition module, and the multi-modal data acquisition module includes a biosensor array unit and a visual acquisition unit; The visual acquisition unit is used to capture visual information of the pet's behavior and appearance and convert it into a digital signal for output; The processing layer includes an edge computing module, and the edge computing module is used for the data transmitted by the biosensor array unit and the visual acquisition unit; The output layer is used to receive the results of the processing layer and perform feedback actions, including an intelligent interaction module, and the intelligent interaction module is used to receive the judgment results of the edge computing unit.

2. The pet health monitoring system based on computer vision and biosensing according to claim 1, characterized in that The edge computing unit includes an integrated spatio-temporal feature fusion model and an anomaly detection algorithm. The integrated spatio-temporal feature fusion model is used to fuse the spatio-temporal features of data from different sources, and the anomaly detection algorithm is used to analyze the processed data to determine whether the pet's health condition is abnormal.

3. The pet health monitoring system based on computer vision and biosensing according to claim 1, wherein The biosensor array unit includes a graphene-based flexible body temperature sensor and a dry electrode type EMG sensor. The graphene-based flexible body temperature sensor is used to accurately measure the pet's body temperature with an accuracy of ±0.1 °C, a response time < 2 s, and transmit the body temperature data. The dry electrode type EMG sensor is used to collect the electromyogram signal of the pet's chewing muscle to judge the pet's eating state and output relevant data.

4. The pet health monitoring system based on computer vision and biosensing according to claim 1, characterized in that, The graphene-based flexible body temperature sensor is made of graphene material to achieve high-precision and fast-response body temperature measurement.

5. The pet health monitoring system based on computer vision and biosensing according to claim 1, characterized in that The dry electrode type EMG sensor can directly collect the electromyogram signal of the pet's chewing muscle without using auxiliary materials such as conductive paste.

6. The pet health monitoring system based on computer vision and biosensing according to claim 2, characterized in that, The visual acquisition unit includes a high-definition camera and an image processor, which are used to capture clear pet visual information and perform preliminary processing.

7. The pet health monitoring system based on computer vision and biosensing according to claim 1, characterized in that, The edge computer module includes a vibration feedback unit and an APP warning push unit. The vibration feedback unit uses a micro vibration motor and can generate vibration signals of different intensities and frequencies. The APP warning push unit is used to push the pet health warning information to the user's mobile APP in real time through network communication technology.

8. A pet health monitoring device based on computer vision and biosensing, such as the pet health monitoring system based on computer vision and biosensing according to claims 1-7, characterized in that, Comprising: An instrument main body, a display screen, a camera, a switch, a charging port, two connecting rings, a strap, and two magic tapes. The display screen is arranged on the surface of the instrument main body, the camera is arranged on the surface of the instrument main body, the switch and the charging port are respectively installed on one side of the surface of the instrument main body, the two connecting rings are symmetrically installed on both sides of the back of the instrument main body, the strap is arranged between the two connecting rings, and the two magic tapes are arranged on both sides of the surface of the strap in a staggered manner.

9. The pet health monitoring device based on computer vision and biosensing according to claim 8, wherein An installation component is arranged between the instrument main body and the connecting ring. The installation component includes an installation sleeve, an installation block, and a fixing bolt. The installation sleeve is connected to the back of the instrument main body, the installation block is connected to the inside of the installation sleeve in a pluggable manner, and the fixing bolt is arranged between the installation sleeve and the installation block.

10. A pet health monitoring method based on computer vision and biosensing, and a pet health monitoring system based on computer vision and biosensing as described in claims 1-7, characterized in that, Including the following steps: S1. Data acquisition stage; S11. Biosensing data acquisition: The flexible body temperature sensor based on graphene exhibits high precision and fast response characteristics. Under the conditions of an accuracy of ±0.1°C and a response time <2 s, it accurately measures the body temperature of the pet and transmits the body temperature data. The dry electrode type EMG sensor directly collects the electromyogram signals of the pet's chewing muscles to judge the pet's eating state and output relevant data; S12. Visual data acquisition: The high-definition camera in the visual acquisition unit captures the visual information of the pet's behavior and appearance, and the image processor converts this visual information into digital signals for output; S2. Data processing stage; S21. Data aggregation: The body temperature, eating state and other data collected by the biosensor array unit, as well as the visual digital signals output by the visual acquisition unit, are all transmitted to the edge computing module in the processing layer; S22. Feature fusion: The integrated spatio-temporal feature fusion model in the edge computing module performs spatio-temporal feature fusion on data from different sources to integrate multi-dimensional information; S23. Health judgment: The anomaly detection algorithm analyzes the data after fusion processing to judge whether the pet's health condition is abnormal; S3. Result feedback stage; S31. Result reception: If the edge computing module judges that the pet's health condition is abnormal, the intelligent interaction module in the output layer receives this judgment result; S32. Vibration feedback: The vibration feedback unit in the intelligent interaction module generates vibration signals with different intensities and frequencies to remind relevant personnel; S33. Early warning push: The APP early warning push unit uses network communication technology to push the pet health early warning information to the user's mobile APP in real time, so that the user can timely understand the pet's health condition.

Images