Medical waste supervision method and device based on AI video engine technology

Abstract

The invention discloses a medical waste supervision method and device based on an AI video engine technology, and relates to the technical field of medical waste supervision, and the method comprises the steps: obtaining appropriate opening regions corresponding to graph nodes in each grid space of an interaction region space, and carrying out the temporary storage of medical wastes, and drawing a real-time opening schematic line of the waste container according to the real-time coordinates of the moving points, comparing the real-time opening schematic line with a proper opening area corresponding to a human hand key node serving as a graph node in a corresponding grid space, judging and triggering an alarm mechanism, and establishing an interaction area space of the human body and the waste container. A lightweight real-time model is used for marking key nodes of hands of a human body and moving points of the waste container to determine a proper opening area corresponding to each grid space, a waste container opening schematic line and the proper opening area are compared in real time, an alarm mechanism is judged and triggered, and the opening state of the waste container is accurately monitored in real time; and risks of medical waste scattering and pathogen transmission are reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical waste supervision, and specifically relates to a medical waste supervision method and device based on AI video engine technology. Background Art

[0002] In medical institutions, the proper treatment of medical waste is of crucial importance. Medical waste contains a large number of bacteria, viruses and harmful substances, which will pose a serious threat to the environment and human health if not treated properly. During the process of medical waste management, temporary storage points or areas are usually set up inside medical institutions, and medical staff regularly collect medical waste centrally.

[0003] However, the supervision of medical waste temporary storage areas mainly relies on manual inspections and video surveillance video playback, and waste containers generally adopt an infrared induction or pressure induction automatic opening design. However, although such induction containers can reduce the risk of operators being contaminated, they cannot determine the matching of the opening state and hand movements in real time. As a result, when the waste container is induction-opened, the induction-opened container may not be fully opened during actual use. The existing supervision methods mainly rely on manual inspections and video surveillance video playback, making it difficult to detect this problem in real time. For example, it may lead to waste spillage and the risk of pathogen spread cannot be contained in time. In addition, simple video surveillance can only record the pictures, lacking the intelligent analysis ability of video content, and the abnormal opening of induction containers is not included in the scope of monitoring, further amplifying the potential safety hazards. Based on this, a medical waste supervision method and device based on AI video engine technology are proposed. Summary of the Invention

[0004] The purpose of the present invention is to provide a medical waste supervision method and device based on AI video engine technology, which solves the technical problem that simple video surveillance can only record pictures and lacks the intelligent analysis ability of video content.

[0005] The medical waste supervision method and device based on AI video engine technology include the following steps: Step 1: Collect video data of interaction events in the medical waste temporary storage area; Step 2: Generate the interaction area space between the human body and the waste container; Step 3: Mark the key nodes of the human hand and the moving points of the waste container located in the medical waste temporary storage area, and then obtain the graph nodes and additional nodes; Step 4: Uniformly divide the interaction area space into multiple identical grid spaces, analyze the position coordinates of the moving points of the waste container when the graph nodes in multiple interaction event video data are in different grid spaces, and obtain the appropriate opening areas corresponding to the graph nodes in each grid space of the interaction area space: Step 5: When temporarily storing medical waste, when the key nodes of the human hand enter the interactive area space, the real-time coordinates of the grid space where the key nodes of the human hand are located and the moving points of the waste container are obtained in real time. According to the real-time coordinates of the moving points, the real-time opening schematic line of the waste container is drawn, and it is compared with the appropriate opening area corresponding to the key nodes of the human hand as graph nodes in the corresponding grid space to determine whether to trigger the alarm mechanism.

[0006] As a further solution of the present invention: The specific method for generating the interactive area space between the human body and the waste container is as follows: Taking the edge of the waste container as a benchmark, expand 0.5 meters outward as the horizontal range, and expand 0.3 meters up and down according to the height of the container in the vertical direction, so as to generate the interactive area space between the human body and the waste container.

[0007] As a further solution of the present invention: The specific method for obtaining graph nodes and additional nodes is as follows: Taking the center point of the top cover of the waste container as the moving point of the waste container; taking the wrist node among the key nodes of the human hand of the graph node as the graph node, and taking the moving point of the waste container as the additional node.

[0008] As a further solution of the present invention: The specific method for obtaining the appropriate opening area corresponding to the graph node in each grid space of the interactive area space is as follows: S1: Select one from each grid space of the interactive area space without replacement as the calibration grid; record the position coordinates of the moving point of the waste container of the graph node in the calibration grid in multiple interactive event video data as Ya (Yxa, Yya, Yza), and the grid positioning point is the center point in each grid space, where a is the position coordinates of the moving point of the waste container in multiple interactive event video data when the graph node is in the calibration grid; S2: Analyze the abscissa, ordinate and vertical coordinate in each moving point position coordinate to obtain the reference abscissa value Hx, reference ordinate Hy and reference vertical coordinate Hz corresponding to the moving point when the graph node is in the calibration grid, obtain the standard deviations Ux, Uy and Uz of the abscissa Yxa, ordinate Yya and vertical coordinate Yza, and take EA (Hx + Ux, Hy + Uy, Hz + Uz) as the upper limit position EA corresponding to the moving point when the graph node is in the calibration grid, and take EB (|Hx - Ux|, |Hy - Uy|, |Hz - Uz|) as the lower limit position EB corresponding to the moving point when the graph node is in the calibration grid, and combine the moving point of the waste container to obtain the appropriate opening area corresponding to the graph node in the calibration grid; S3: Repeat steps S1 - S2 to obtain the appropriate opening area corresponding to the graph node in each grid space of the interactive area space.

[0009] As a further solution of the present invention: The specific method for obtaining the reference abscissa value, reference ordinate, and reference vertical coordinate corresponding to the moving point when the graph node is in the calibration grid is as follows: Obtain the number v of values in the abscissa Yxa of the position coordinates of each moving point that satisfy |Yxa - Yxp| ≥ Y2. When the number v is greater than the preset threshold Y1, define the mean value Yxp of Yxa as the reference abscissa value Hx corresponding to the moving point when the graph node is in the calibration grid. When the number v is less than the preset threshold Y1, define the mean value of the maximum and minimum values in Yxa as the reference abscissa value Hx corresponding to the moving point when the graph node is in the calibration grid; Use the same analysis method to analyze the ordinate Yya and vertical coordinate Yza of the position coordinates of each moving point, and then obtain the reference ordinate Hy and reference vertical coordinate Hz corresponding to the moving point when the graph node is in the calibration grid.

[0010] As a further solution of the present invention: Combine the moving points of the waste container to obtain the appropriate opening area corresponding to the graph node in the calibration grid: Draw a perpendicular line from the moving point of the waste container to the axis of rotation of the top cover of the waste container, and use the intersection point between the perpendicular line and the axis of rotation as the rotation axis point Z. Connect the upper limit point EA and the lower limit point EB corresponding to the moving point when the graph node is in the calibration grid to the rotation axis point Z respectively, and at the same time connect the upper limit point EA and the lower limit point EB, so as to obtain the appropriate opening area corresponding to the graph node in the calibration grid.

[0011] As a further solution of the present invention: The specific method for determining to trigger the alarm mechanism is as follows: Determine whether the real-time opening schematic line is within the corresponding appropriate opening area. If it is, do nothing; if not, continuously obtain the key nodes of the human hand e times as the real-time opening schematic lines corresponding to the graph node in the interaction area space, and obtain the number g of the real-time opening schematic lines that are within the corresponding appropriate opening area. When g is greater than the preset value Y3, do nothing. When g is less than or equal to the preset value Y3, trigger the alarm mechanism, and the value of e is 4.

[0012] The medical waste supervision device based on the AI video engine technology includes: A data acquisition end, which acquires the video data of the interaction event in the medical waste temporary storage area; An interaction area space generation end, which generates the interaction area space between the human body and the waste container; A graph node and additional node marking end, which marks the key nodes of the human hand and the moving points of the waste container in the medical waste temporary storage area to obtain the graph node and the additional node; The moderate opening area acquisition end evenly divides the interactive area space into multiple identical grid spaces, and obtains the corresponding moderate opening areas when the graph nodes are in each grid space of the interactive area space: The alarm mechanism trigger determination end, when temporarily storing medical waste, when the key nodes of the human hand enter the interactive area space, it obtains the grid space where the key nodes of the human hand are located and the real-time coordinates of the moving points of the waste container in real time, draws the real-time opening schematic line of the waste container, and compares it with the moderate opening area corresponding to the graph nodes of the key nodes of the human hand in the corresponding grid space to determine whether to trigger the alarm mechanism.

[0013] Compared with the prior art, the beneficial effects of the present invention are: (1) In the present invention, through the collection of interactive event video data in the medical waste temporary storage area and the accurate marking of the key nodes of the human hand and the moving points of the waste container, combined with the grid division and data analysis of the interactive area space, it can accurately judge the moderate opening area of the waste container in different positions, realize the accurate monitoring and judgment of the opening state of the waste container, and greatly improve the accuracy and reliability of supervision compared with the traditional supervision method; (2) In the present invention, when the real-time opening schematic line of the waste container is not in the corresponding moderate opening area, through continuous multiple judgments and statistics, it decides whether to trigger an alarm, which can effectively reduce false alarms; (3) In the present invention, by constructing the interactive area space between the human body and the waste container, and using a lightweight real-time model to mark the key nodes of the human hand and the moving points of the waste container to determine the corresponding moderate opening area of each grid space, when the human hand enters the interactive area, it compares the real-time opening schematic line of the waste container with the moderate opening area in real time to determine whether to trigger the alarm mechanism, effectively overcoming the lag of the existing supervision means, realizing the real-time and accurate monitoring of the opening state of the waste container, being able to timely warn of abnormal situations, reducing the risk of medical waste spillage and pathogen transmission, and at the same time improving the supervision efficiency, providing a strong guarantee for the standardized and intelligent supervision of medical waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic structural diagram of the system framework of the present invention; Figure 2 It is a schematic structural diagram of the device of the present invention; Figure 3 It is a schematic structural diagram of the moderate opening area D1 of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0015] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0016] Embodiment 1: Please refer to Figure 1 and Figure 3 , the present application provides a medical waste supervision method based on AI video engine technology, including the following steps: Step 1: Deploy multi-angle cameras inside the medical waste temporary storage area to achieve full coverage of the medical waste temporary storage area, and collect video data of interaction events in the medical waste temporary storage area. The video data of interaction events refers to the video data when medical staff put medical waste. Reasonably deploy multi-angle cameras in the medical waste temporary storage area to achieve full coverage of the area, and focus on collecting video data of interaction events such as medical staff putting medical waste, providing basic information for subsequent analysis. Step 2: Obtain the interactive area space between the human body and the waste container. The specific method is as follows: Taking the edge of the waste container as the benchmark, expand 0.5 meters outward as the horizontal range, and expand 0.3 meters up and down according to the height of the container in the vertical direction, thereby forming a three-dimensional interactive area space, and then generating the interactive area space between the human body and the waste container.

[0017] Step 3: Use a lightweight real-time model, such as the MediaPipe Pose or HRNet model, to mark the key nodes of the human hand and the moving points of the waste containers located in the medical waste temporary storage area, and then obtain graph nodes and additional nodes. The specific method is as follows: Take the center point of the top cover of the waste container as the moving point of the waste container; take the wrist node in the key nodes of the human hand of the graph node as the graph node, and take the position point and moving point of the waste container as the additional nodes; Step 4: Evenly divide the interactive area space into multiple identical grid spaces, analyze the position coordinates of the moving points of the waste containers when the graph nodes in the video data of multiple interaction events are in different grid spaces, and obtain the appropriate opening areas corresponding to the graph nodes in each grid space of the interactive area space. The specific method is as follows: S1: Select one grid space from each grid space in the interaction area space without replacement as the calibration grid; record the moving point position coordinates of the graph node in the waste container in the calibration grid in multiple interactive event video data as Ya (Yxa, Yya, Yza), and the grid positioning point is the center point in each grid space, where a is the moving point position coordinates of the waste container in multiple interactive event video data when the graph node is in the calibration grid; S2: Obtain the value Yxv that satisfies the preset screening condition AI among the abscissas Yxa of each moving point position coordinate, where v is the number of values that satisfy the preset screening condition AI in Yxa. Compare the number v with the preset threshold Y1. When the number v is greater than the preset threshold Y1, it indicates that the number of values that satisfy the preset screening condition AI in the abscissa Yxa is relatively large, and the mean value of Yxa is representative. Then, define the mean value Yxp of Yxa as the reference abscissa value Hx corresponding to the moving point when the graph node is in the calibration grid. When the number v is less than the preset threshold Y1, it indicates that the number of values that satisfy the preset screening condition AI in Yxa is relatively small, and the mean value of Yxa is not representative. Then, define the mean value of the maximum and minimum values in Yxa as the reference abscissa value Hx corresponding to the moving point when the graph node is in the calibration grid; Here, the preset condition AI is specifically: |Yxa - Yxp| ≥ Y2, where Y2 is a preset value, and the specific values of Y1 and Y2 are both determined by relevant personnel according to actual needs; Analyze the ordinates Yya and vertical coordinates Yza of each moving point position coordinate in the same analysis method, and then obtain the reference ordinate Hy and reference vertical coordinate Hz corresponding to the moving point when the graph node is in the calibration grid; Obtain the standard deviations Ux, Uy, and Uz of the abscissa Yxa, ordinate Yya, and vertical coordinate Yza. Take EA (Hx + Ux, Hy + Uy, Hz + Uz) as the upper limit position EA corresponding to the moving point when the graph node is in the calibration grid, and take EB (|Hx - Ux|, |Hy - Uy|, |Hz - Uz|) as the lower limit position EB corresponding to the moving point when the graph node is in the calibration grid; Draw a perpendicular line from the moving point of the waste container to the axis of rotation of the top cover of the object container, and take the intersection point between the perpendicular line and the axis of rotation as the rotation axis point Z. Connect the upper limit position EA and the lower limit position EB corresponding to the moving point when the graph node is in the calibration grid to the rotation axis point Z respectively, and at the same time connect the upper limit position EA and the lower limit position EB, and then obtain the appropriate opening area D1 corresponding to the graph node in the calibration grid; S3: Repeat steps S1 - S2, and the appropriate opening area Dr corresponding to the graph node in each grid space of the interaction area space can be obtained, where r represents different grid spaces in the interaction area space; By collecting video data of interaction events in the temporary storage area of medical waste and accurately marking the key nodes of the human hand and the moving points of waste containers, combined with grid division and data analysis of the interaction area space, it is possible to accurately judge the appropriate opening area of the waste container at different positions, realize accurate monitoring and judgment of the opening state of the waste container, and greatly improve the accuracy and reliability of supervision compared with traditional supervision methods.

[0018] Step Five: When temporarily storing medical waste, when the key nodes of the human hand enter the interaction area space, the grid space where the key nodes of the human hand are located is obtained in real time, and then the corresponding grid space where the graph nodes are located is obtained. The real-time coordinates of the moving points of the waste container are obtained in real time. According to the real-time coordinates of the moving points, a real-time opening indication line of the waste container is drawn. The real-time opening indication line of the waste container is compared with the appropriate opening area corresponding to the graph nodes where the key nodes of the human hand are located in the corresponding grid space to determine whether the real-time opening indication line is within the corresponding appropriate opening area. If it is, no processing is performed. If it is not, for e consecutive times, according to the real-time opening indication lines corresponding to the key nodes of the human hand as graph nodes in the interaction area space, the number g of the real-time opening indication lines that are within the corresponding appropriate opening area is obtained. When g is greater than the preset value Y3, no processing is performed. When g is less than or equal to the preset value Y3, the alarm mechanism is triggered to remind the person putting in the waste that there is an abnormality in the opening of the waste container, and at the same time, an abnormal message is sent to the supervisor. The specific values of the preset value Y3 are all determined by relevant personnel according to actual needs, and the value of e is 4. An alarm mechanism based on real-time coordinate comparison and statistical analysis is established. When the real-time opening indication line of the waste container is not within the corresponding appropriate opening area, through continuous multiple judgments and statistics, it is decided whether to trigger the alarm, which can effectively reduce false alarms and timely remind the person putting in the waste and the supervisor.

[0019] When an abnormal opening state of the waste container is detected, the alarm can be triggered in a timely manner according to the set alarm mechanism to remind the person putting in the waste and the supervisor, so that timely measures can be taken for treatment, effectively avoiding risks such as medical waste spillage and germ transmission caused by improper opening of the waste container, ensuring the safety and standardization of the medical waste putting-in link, and at the same time realizing accurate supervision of the opening state of the container during the medical waste putting-in process.

[0020] Embodiment Two: Please refer to Figure 2 , as Embodiment Two of the present invention, a medical waste supervision device based on AI video engine technology is provided. This device is used to apply the control method of a solid-state drive disclosed above, and specifically includes; Data acquisition terminal: Deploy multi-angle cameras inside the medical waste temporary storage area to achieve full coverage of the medical waste temporary storage area and collect video data of interaction events in the medical waste temporary storage area. The video data of interaction events refers to the video data when medical staff put medical waste. Interactive area space generation terminal: Taking the edge of the waste container as a benchmark, expand 0.5 meters outward as the horizontal range, and expand 0.3 meters up and down according to the height of the container in the vertical direction, so as to form a three-dimensional interactive area space, and then generate the interactive area space between the human body and the waste container. Graph node and additional node marking terminal: Take the center point of the waste container top cover as the moving point of the waste container; Take the wrist node in the key nodes of the human hand in the graph node as the graph node, and take the position point and moving point of the waste container as the additional nodes. Moderate opening area acquisition terminal: Divide the interactive area space evenly into multiple identical grid spaces, and obtain the corresponding moderate opening areas when the graph node is in each grid space of the interactive area space: Alarm mechanism trigger determination terminal: When storing medical waste, when the key nodes of the human hand enter the interactive area space, obtain the grid space where the key nodes of the human hand are located in real time, and the real-time coordinates of the moving point of the waste container, draw the real-time opening schematic line of the waste container, and compare it with the moderate opening area corresponding to the graph node (the key nodes of the human hand) in the corresponding grid space to determine whether to trigger the alarm mechanism. When storing medical waste, obtain the grid space where the key nodes of the human hand are located in real time, and then obtain the corresponding grid space where the graph node is located, obtain the real-time coordinates of the moving point of the waste container in real time, draw the real-time opening schematic line of the waste container according to the real-time coordinates of the moving point, and compare the real-time opening schematic line of the waste container with the moderate opening area corresponding to the graph node (the key nodes of the human hand) in the corresponding grid space to determine whether the real-time opening schematic line is within the corresponding moderate opening area. If it is, do nothing. If not, continuously obtain the number g of the real-time opening schematic lines that are within the corresponding moderate opening area among the real-time opening schematic lines corresponding to the graph node (the key nodes of the human hand) in the interactive area space e times. When g is greater than the preset value Y3, do nothing. When g is less than or equal to the preset value Y3, trigger the alarm mechanism to remind the person putting the waste that there is an abnormality in the opening of the waste container, and at the same time send an abnormal message to the supervisor. The specific value of the preset value Y3 is determined by relevant personnel according to actual needs, and the value of e is 4. Video data of the disposal behavior in the medical waste temporary storage area is collected through a multi-angle camera, an interaction area space between the human body and the waste container is constructed, and key nodes of the human hand and the moving points of the waste container are marked using a lightweight real-time model; after grid division and data screening and analysis, the appropriate opening area corresponding to each grid space is determined. When the human hand enters the interaction area, the opening indication line of the waste container is compared with the appropriate opening area in real time, and the alarm mechanism is triggered through multiple judgments. This solution effectively overcomes the lag and inaccuracy of existing supervision means, realizes real-time and accurate monitoring of the opening state of the waste container, can timely warn of abnormal situations, reduce the risk of medical waste spillage and pathogen transmission, while reducing manual intervention, improving supervision efficiency, and forming traceable data records, providing a strong guarantee for the standardized and intelligent supervision of medical waste.

[0021] Embodiment 3: As Embodiment 3 of the present invention, in the specific implementation of this application, compared with Embodiment 1 and Embodiment 2, the technical solution of this embodiment lies in combining the solutions of the above Embodiment 1 and Embodiment 2 for implementation.

[0022] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0023] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. A medical waste supervision method based on AI video engine technology, characterized in that: The following steps are involved: Step 1: Collect video data of interactive events in the temporary storage area for medical waste; Step 2: Generate the interaction area space between the human body and the waste container; Step 3: Mark the key nodes of the human hand and the moving points of the waste container in the temporary storage area of ​​medical waste, and then obtain the graph nodes and additional nodes; Step 4: Evenly divide the interactive area space into multiple identical grid spaces, analyze the moving point position coordinates of the waste container when the graph nodes in multiple interactive event video data are in different grid spaces, and obtain the appropriate opening areas corresponding to the graph nodes in each grid space of the interactive area space: Step 5: When medical waste is temporarily stored, when the key node of the human hand enters the interactive area space, the real-time coordinates of the grid space where the key node of the human hand is located and the moving point of the waste container are obtained in real time, and the real-time opening schematic line of the waste container is drawn according to the real-time coordinates of the moving point. It is compared with the appropriate opening area corresponding to the key node of the human hand as a graph node in the corresponding grid space to determine whether to trigger the alarm mechanism.

2. The medical waste supervision method based on AI video engine technology according to claim 1 is characterized in that: The specific method of generating the interaction area space between the human body and the waste container is: Taking the edge of the waste container as the reference, the horizontal range is expanded outward by 0.5 meters, and the vertical range is expanded up and down by 0.3 meters according to the height of the container, thereby generating an interactive area space between the human body and the waste container.

3. The medical waste supervision method based on AI video engine technology according to claim 1 is characterized in that: The specific way to obtain graph nodes and additional nodes is: The center point of the waste container top cover is used as the moving point of the waste container; the wrist node among the key nodes of the human hand in the graph node is used as the graph node, and the moving point of the waste container is used as the additional node.

4. The medical waste supervision method based on AI video engine technology according to claim 3 is characterized in that: The specific method of obtaining the appropriate opening areas corresponding to the graph nodes in each grid space of the interactive area space is as follows: S1: Select one of the grid spaces in the interactive area space as a calibration grid without replacement; mark the moving point position coordinates of the waste container in the calibration grid of the graph node in the multiple interactive event video data as Ya (Yxa, Yya, Yza), the grid positioning point is the center point in each grid space, where a is the moving point position coordinates of the waste container in the multiple interactive event video data when the graph node is in the calibration grid; S2: Analyze the horizontal coordinate, vertical coordinate and vertical coordinate in the position coordinate of each moving point, obtain the reference horizontal coordinate value Hx, reference vertical coordinate Hy and reference vertical coordinate Hz corresponding to the moving point when the graph node is in the calibration grid, obtain the standard deviation Ux, Uy and Uz of the horizontal coordinate Yxa, vertical coordinate Yya and vertical coordinate Yza, take EA (Hx + Ux, Hy + Uy, Hz + Uz) as the upper limit point EA corresponding to the moving point when the graph node is in the calibration grid, take EB (|Hx-Ux|, |Hy-Uy|, |Hz-Uz|) as the lower limit point EB corresponding to the moving point when the graph node is in the calibration grid, and obtain the appropriate opening area corresponding to the graph node in the calibration grid in combination with the moving point of the waste container; S3: Repeat steps S1-S2 to obtain appropriate opening areas corresponding to graph nodes in each grid space of the interaction area space.

5. The medical waste supervision method based on AI video engine technology according to claim 4 is characterized in that: The specific method for obtaining the reference abscissa value, reference ordinate value, and reference vertical coordinate value corresponding to the moving point of the graph node in the calibration grid is: Obtain the number v of values ​​satisfying |Yxa-Yxp|≥Y2 in the horizontal coordinate Yxa of the position coordinates of each moving point. When the number v is greater than the preset threshold value Y1, define the mean value Yxp of Yxa as the reference horizontal coordinate value Hx corresponding to the moving point when the graph node is in the calibration grid. When the number v is less than the preset threshold value Y1, define the mean of the maximum and minimum values ​​in Yxa as the reference horizontal coordinate value Hx corresponding to the moving point when the graph node is in the calibration grid. The same analysis method is used to analyze the ordinate Yya and vertical coordinate Yza of the position coordinates of each moving point, and then the reference ordinate Hy and reference vertical coordinate Hz corresponding to the moving point when the graph node is in the calibration grid are obtained.

6. The medical waste supervision method based on AI video engine technology according to claim 4 is characterized in that: Combine the moving points of the waste container to obtain the appropriate opening area corresponding to the graph node in the calibration grid: Draw a perpendicular line from the moving point of the waste container to the rotation axis of the top cover of the waste container, and take the intersection of the perpendicular line and the rotation axis as the rotation axis point Z. Connect the upper limit point EA and the lower limit point EB corresponding to the moving point of the graph node in the calibration grid to the rotation axis point Z respectively, and connect the upper limit point EA and the lower limit point EB at the same time, so as to obtain the appropriate opening area corresponding to the graph node in the calibration grid.

7. The medical waste supervision method based on AI video engine technology according to claim 6 is characterized in that: The specific method for determining whether to trigger the alarm mechanism is: Determine whether the real-time opening schematic line is in the corresponding appropriate opening area. If so, do not do anything. If not, obtain the real-time opening schematic lines corresponding to the key nodes of the human hand as graph nodes in the interactive area space for e consecutive times, and obtain the number g of each real-time opening schematic line in the corresponding appropriate opening area. When g is greater than the preset value Y3, do not do anything. When g is less than or equal to the preset value Y3, trigger the alarm mechanism, and the value of e is 4.

8. The medical waste monitoring device based on AI video engine technology is characterized by: The device applies the medical waste supervision method based on AI video engine technology described in any one of claims 1 to 7, including: The data collection end collects video data of interactive events in the temporary storage area for medical waste; An interactive area space generating end generates an interactive area space between a human body and a waste container; A graph node and additional node marking end marks key nodes of the human hand and moving points of waste containers in the temporary storage area for medical waste to obtain graph nodes and additional nodes; The appropriate opening area acquisition end evenly divides the interactive area space into multiple identical grid spaces, and obtains the appropriate opening areas corresponding to the graph nodes in each grid space of the interactive area space: At the alarm mechanism triggering judgment end, when medical waste is temporarily stored, when the key node of the human hand enters the interactive area space, the grid space where the key node of the human hand is located and the real-time coordinates of the moving point of the waste container are obtained in real time to draw the real-time opening schematic line of the waste container, and compare it with the appropriate opening area corresponding to the key node of the human hand as a graph node in the corresponding grid space to determine the triggering of the alarm mechanism.

Images