A pest control map generation system based on computer vision

Abstract

This invention relates to the field of computer vision technology, and more particularly to a computer vision-based pest control map generation system. The system includes: an image acquisition module for acquiring video streams from a camera and extracting images frame by frame; a target recognition module for detecting targets in the extracted images using a target detection model to obtain the confidence level of each target and the bounding box of valid targets; a mapping analysis module for constructing the actual ground coordinates of valid targets; a grid division module for dividing the targets; a map generation module for classifying the target heat level of each grid and generating a pest control map based on the target heat level classification results of each grid within a management period; and a confidence analysis module for adjusting the pest control map generation process based on the confidence level of valid targets in each grid within a management period. This invention improves pest control efficiency.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular to a pest control map generation system based on computer vision. Background Art

[0002] Traditional methods of catching and disinfecting cockroaches mostly rely on manual inspections, which is not only time-consuming and laborious, but also easily affected by subjective factors, resulting in inaccurate monitoring results and greatly reduced pest control effects. In recent years, with the rapid development of computer vision, artificial intelligence, and deep learning technologies, automated pest detection and management systems based on visual recognition have emerged. These systems can monitor targets in real time, identify different types of pests, and generate corresponding pest control maps through data analysis, providing a scientific basis for pest control.

[0003] Chinese patent publication number CN116251214A discloses a full-process disinfection monitoring method for an intelligent disinfection robot. The intelligent disinfection robot includes a mobile platform, a control processing system, a disinfection component, and a human body detection sensor group. The full-process disinfection monitoring method is adopted: after the real-time disinfection mode ends, it switches to the patrol monitoring mode. In the whole disinfection process, the human body detection sensor group and the human body data detection module are used to analyze whether there are people in the surrounding environment: if there are people, in the real-time disinfection mode, a signal is sent to the control processing system to suspend or close the disinfection component, and in the patrol monitoring mode, a signal is sent to the control processing system to suspend patrol and prompt the personnel to leave. The present invention realizes complete safety disinfection management through patrol monitoring of real-time disinfection and the dangerous period of the scene after disinfection; further, the exposed personnel recording module is used to timely protect the relevant exposed personnel, and the recognition rate and disinfection efficiency of people are improved by combining and analyzing the data of the human body detection sensor group; it can be seen that the scheme does not analyze the disinfection data, does not realize the dynamic learning adjustment of disinfection, and has the problem of low disinfection efficiency. Summary of the invention

[0004] The object of the present invention is to provide a pest control map generation system based on computer vision to solve at least one of the problems existing in the prior art.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A computer vision-based disinfection map generation system, comprising:

[0007] The image acquisition module is used to collect the video stream of the camera and perform frame processing on the video stream to extract the image;

[0008] The target recognition module is used to perform target detection on the extracted image according to the target detection model to obtain the confidence of each target, and is also used to analyze the validity of the target according to the confidence of each target to obtain the bounding box of the valid target;

[0009] A mapping analysis module is used to construct the actual ground coordinates of the valid target according to the border of the valid target, the camera installation height and the camera focal length;

[0010] A grid division module is used to grid the ground and divide the target according to the construction results of the actual ground coordinates of the valid target;

[0011] A map generation module is used to divide the target heat level of each grid according to the target division results in the monitoring period, and is also used to generate a disinfection map according to the division results of the target heat level of each grid in the management period, and output the generated disinfection map to the user;

[0012] The confidence analysis module is used to adjust the generation process of the disinfection map according to the confidence of the effective targets of each grid within the management cycle.

[0013] Furthermore, the target recognition module is used to perform target detection on the extracted image according to the target detection model to obtain the confidence Zi of the i-th target in the image, and compare the confidence Zi of the i-th target in the image with the preset confidence threshold z0 to analyze the validity of the target. The validity of the target includes valid targets and invalid targets. If the i-th target is a valid target, it is taken as the i valid target, and the border of the i valid target is set to Ki(xi_center,yi_center,wi,hi), xi_center is the horizontal coordinate of the center point of the i valid target border, yi_center is the vertical coordinate of the center point of the i valid target border, wi is the width of the i valid target bounding box, and hi is the height of the i valid target bounding box.

[0014] Furthermore, the mapping analysis module is used to construct the actual ground coordinates of the effective target according to the frame of the effective target, the camera installation height c and the camera focal length f, so as to obtain the actual ground coordinates (Sxi, Syi) of the center point of the i effective target.

[0015] Furthermore, the grid division module includes a grid division unit, which is used to divide the long side of the circumscribed rectangle of the ground into n equal parts, and divide the short side of the circumscribed rectangle of the ground into m equal parts, so as to divide the circumscribed rectangle of the ground into m×n grids.

[0016] Furthermore, the grid division module further includes a target division unit, wherein the target division unit is used to divide the target according to the grid division result and the construction result of the actual ground coordinates of the valid target, wherein:

[0017] If (Sxi,Syi)∈U jk {x, y|(j-1)×L1 / n≤x<j×L1 / n, (k-1)×L2 / m≤y<k×L2 / m}, the target partitioning unit partitions the i valid targets into U jk Otherwise, the target partitioning unit does not partition i valid targets into U jk , where L1 is the length of the long side of the circumscribed rectangle of the ground, L2 is the length of the short side of the circumscribed rectangle of the ground, 0<j≤n, 0<k≤m, U jk It is the area of ​​the grid whose long side is j and short side is k in the circumscribed rectangle of the ground.

[0018] Furthermore, the map generation module includes a level division unit, which is used to divide the target heat level of each grid according to the target division result within the monitoring period, and the target heat level includes level one, level two and level three.

[0019] Furthermore, the map generation module further includes a region division unit, which is used to divide each grid type according to the division result of the target heat level of each grid in each monitoring period within the management period, wherein:

[0020] If (w1×sjk1+w2×sjk2) / S=0, the area division unit determines that the grid type of the grid whose current management period area is Ujk is a safe area;

[0021] If (w1×sjk1+w2×sjk2) / S≤T, the area division unit determines that the grid type of the grid whose current management period area is Ujk is a medium-risk area;

[0022] If (w1×sjk1+w2×sjk2) / S>T, the area division unit determines that the grid type of the grid whose current management period area is Ujk is a high-risk area;

[0023] Among them, w1 is the weight of the grid with target heat level of level 2, w2 is the weight of the grid with target heat level of level 3, sjk1 is the weight of the grid with area U in the management period jk The number of monitoring cycles with grid target heat level of level 2, sjk2 is the area U within the management period jk The number of monitoring cycles with grid target heat level 3, S is the number of monitoring cycles in the management period, and T is the preset proportion threshold.

[0024] Furthermore, the confidence analysis module includes a confidence analysis unit, which is used to analyze the abnormality of the confidence according to the confidence of each grid valid target within the management period. The abnormality of the confidence includes normal and abnormal. If the area is U jk The confidence level of the grid valid target is abnormal, and the preset abnormal proportion threshold is set to T1.

[0025] Furthermore, the confidence analysis module further includes a construction unit, which is used to determine the confidence level of the area according to the management period. jk The confidence of the effective target in the grid is used to construct the fluctuation coefficient of the confidence to obtain the area U within the management period. jk The fluctuation coefficient BA of the confidence of valid targets in the grid.

[0026] Furthermore, the confidence analysis module further includes a fluctuation analysis unit, which manages the area U within the period jk The fluctuation coefficient BA of the confidence of the effective target in the grid is compared with the preset fluctuation coefficient B0, and the volatility of the confidence is analyzed according to the comparison result. The volatility of the confidence includes normal fluctuation and abnormal fluctuation. If the area in the current management cycle is U jk The fluctuation of the confidence of valid targets in the grid is abnormal fluctuation, and the preset threshold is adjusted to ZX1.

[0027] The beneficial effects of the present invention are as follows: the system significantly improves the efficiency and accuracy of cockroach extermination operations by integrating advanced image acquisition, target detection and geographic information processing technologies. Its image acquisition module ensures high-quality video streaming, and through the mapping analysis module, the system can convert the captured target position into actual ground coordinates to ensure the accuracy of positioning. In addition, the system uses grid division to visualize the triggered extermination needs as a heat map, allowing managers to quickly identify high-risk areas and thus give priority to targeted extermination. The intelligent setting of the monitoring and management cycle realizes dynamic data tracking and feedback, and improves the scientific nature of the extermination measures. In addition, the confidence and fluctuation analysis mechanism enhances the system's adaptive adjustment capabilities and ensures the objectivity and stability of the data. Overall, the system improves the level of intelligence in pest management and provides users with an efficient, intuitive and reliable extermination solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0029] Figure 1Schematic diagram of the structure of the computer vision-based disinfection map generation system of this embodiment.

[0030] Figure 2 Schematic diagram of the structure of the grid division module of this embodiment.

[0031] Figure 3 This is a schematic diagram of the structure of the map generation module in this embodiment.

[0032] Figure 4 Schematic diagram of the structure of the confidence analysis module of this embodiment. DETAILED DESCRIPTION

[0033] In order to more clearly illustrate the present invention, the present invention is further described below in conjunction with preferred embodiments and accompanying drawings. Similar components in the accompanying drawings are represented by the same reference numerals. It should be understood by those skilled in the art that the content specifically described below is illustrative rather than restrictive, and should not be used to limit the scope of protection of the present invention.

[0034] It should be noted that, although the terms first, second, third, etc. may be used to describe in the embodiments of the present application, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, without departing from the scope of the embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first.

[0035] See also Figure 1 As shown, it is a schematic diagram of the structure of the disinfecting map generation system based on computer vision in this embodiment, and the system includes:

[0036] The image acquisition module is used to acquire the video stream of the camera and perform frame processing on the video stream to extract images. It can be understood that the present embodiment does not specifically limit the method of acquiring the video stream of the camera and extracting images by framing. Those skilled in the art can freely set it as long as the requirements of acquiring the video stream of the camera and extracting images by framing are met. Among them, the video stream can be captured from the camera through OpenCV and images can be extracted ten times per second.

[0037] Please continue reading Figure 1As shown, the system also includes a target recognition module, which is connected to the image acquisition module. The target recognition module is used to perform target detection on the extracted image according to the target detection model to obtain the confidence of each target. The target recognition module sets the confidence of the i-th target in the image to Zi, and compares the confidence Zi of the i-th target in the image with the preset confidence threshold z0 to analyze the validity of the target. If Zi≤z0, the target recognition module determines that the i-th target is an invalid target. If Zi>z0, the target recognition module determines that the i-th target is a valid target and uses it as the i-th valid target. The target recognition module sets the border of the i-th valid target to Ki(xi_center,yi_center,wi,hi), where (xi_center,yi_center) is the coordinate of the center point of the border of the i-th valid target, xi_center is the horizontal coordinate of the center point of the border of the i-th valid target, and y i_center is the ordinate of the center point of the i valid target bounding box, wi is the width of the i valid target bounding box, and hi is the height of the i valid target bounding box. The model can realize real-time target detection, which is convenient for cockroach detection in practical applications. By setting the confidence threshold, it can effectively filter out misidentified and low-confidence targets, ensuring that subsequent processing is based on high-quality data. For pest control, it helps to improve the accuracy of pest control map generation.

[0038] It is understandable that the setting of the preset confidence threshold is not specifically limited in this embodiment, and those skilled in the art can freely set it as long as the setting requirements of the preset confidence threshold are met, wherein the optimal value of z0 is 0.5.

[0039] Specifically, the borders of the i valid targets described in this embodiment are coordinates in the pixel coordinate system of the image, with the upper left corner of the image as the origin, the (x) coordinate increasing from left to right, and the (y) coordinate increasing from top to bottom.

[0040] Specifically, the target described in this embodiment is cockroach.

[0041] It can be understood that the setting of the target detection model is not specifically limited in this embodiment, and those skilled in the art can set it freely as long as the target detection requirements are met. Among them, the lightweight version of YOLOv5, YOLOv5s, can be used as the basic model for target detection, and the target detection model is trained using the insect detection dataset. Through the training data format, the targets and their normalized coordinates marked in each frame are annotated, and data enhancement and virtual data expansion are performed. The model output results are the target bounding box position and the target confidence.

[0042] Please continue reading Figure 1As shown, the system also includes a mapping analysis module, which is connected to the target recognition module. The mapping analysis module is used to construct the actual ground coordinates of the effective target according to the frame of the effective target, the camera installation height c and the camera focal length f, so as to accurately locate the position of the cockroach in the actual environment. The mapping analysis module sets the actual ground coordinates of the center point of the i effective target to (Sx i, Sy i), and sets:

[0043] Sx i=(x i_center-u0)×c / f;

[0044] Sy i=(y i_center-v0)×c / f;

[0045] Among them, u0 is the horizontal coordinate of the principal point of the camera image coordinate system, and v0 is the vertical coordinate of the principal point of the camera image coordinate system. By mapping the target from pixel coordinates to actual ground coordinates, the position of cockroaches in the actual environment can be accurately located. The appropriate focal length and installation height ensure that the field of view and height of the monitoring area are adapted to avoid blind spots.

[0046] Specifically, the actual ground coordinates of the effective target described in this embodiment are a plane rectangular coordinate system on the indoor ground, which is established with the lower left corner of the ground circumscribed rectangle as the coordinate origin, the long side of the circumscribed rectangle as the x-axis, and the short side of the circumscribed rectangle as the y-axis.

[0047] It can be understood that the present embodiment does not specifically limit the method for obtaining the camera installation height and the camera focal length, and those skilled in the art can freely set it as long as the requirements for obtaining the camera installation height and the camera focal length are met, wherein the camera installation height and the camera focal length can be obtained through interaction.

[0048] It can be understood that, in this embodiment, there is no specific limitation on the method for obtaining the horizontal coordinate of the principal point of the camera image coordinate system and the vertical coordinate of the principal point of the camera image coordinate system. Those skilled in the art can set it freely, and it is only necessary to meet the requirements for obtaining the horizontal coordinate of the principal point of the camera image coordinate system and the vertical coordinate of the principal point of the camera image coordinate system. Among them, the position of the principal point of the camera image coordinate system can be obtained by performing camera calibration through OpenCV.

[0049] Please continue reading Figure 1 As shown, the system further includes a grid division module, which is connected to the mapping analysis module, and the grid division module is used to grid the ground and divide the target according to the construction result of the actual ground coordinates of the valid target.

[0050] See also Figure 2As shown, the grid division module includes a grid division unit, which is used to divide the long side of the circumscribed rectangle of the ground into n equal parts, and divide the short side of the circumscribed rectangle of the ground into m equal parts, so as to divide the circumscribed rectangle of the ground into m×n grids; using uniform grid division, the distribution of cockroaches in each area can be accurately evaluated, which is convenient for subsequent pest control operations to be performed, and can make the pest control work more targeted.

[0051] It is understandable that the present embodiment does not specifically limit the settings of m and n, and those skilled in the art may freely set them as long as the setting requirements of m and n are met, wherein the optimal value of m is 25, and the optimal value of n is 30.

[0052] Please continue reading Figure 2 As shown, the grid division module also includes a target division unit, which is connected to the grid division unit, and the target division unit is used to divide the target according to the grid division result and the construction result of the actual ground coordinates of the valid target, so as to accurately divide the detected valid target into the corresponding grid according to the actual ground coordinates, wherein:

[0053] If (Sxi,Syi)∈U jk {x, y|(j-1)×L1 / n≤x<j×L1 / n, (k-1)×L2 / m≤y<k×L2 / m}, the target partitioning unit partitions the i valid targets into U jk Otherwise, the target partitioning unit does not partition i valid targets into U jk , where L1 is the length of the long side of the circumscribed rectangle of the ground, L2 is the length of the short side of the circumscribed rectangle of the ground, 0<j≤n, 0<k≤m, U jk It is the area of ​​the grid with the jth long side and the kth short side in the circumscribed rectangle of the ground. Through the distribution of effective targets in the grid, we can intuitively understand the distribution characteristics of pests, help formulate subsequent pest control and prevention measures, and improve the accuracy of pest control map generation.

[0054] Specifically, the length of the long side of the circumscribed rectangle of the ground and the length of the short side of the circumscribed rectangle of the ground in this embodiment can be obtained through interaction.

[0055] Please continue reading Figure 1As shown, the system also includes a map generation module, which is connected to the grid division module, and the map generation module is used to divide the target heat level of each grid according to the division result of the target within the monitoring period, and is also used to generate a disinfection map according to the division result of the target heat level of each grid within the management period, and output the disinfection map to the user; it can be understood that the present embodiment does not make specific limitations on the setting of the monitoring period and the management period, and those skilled in the art can set them freely, and only need to meet the setting requirements of the monitoring period and the management period, wherein the setting of the management period should be greater than the setting of the monitoring period, the management period can be set to 7 days, and the monitoring period can be set to 1h.

[0056] See also Figure 3 As shown, the map generation module includes a level division unit, which is used to divide the target heat level of each grid according to the division result of the target within the monitoring period, so as to express the pest degree of a certain grid in a result-oriented manner, wherein:

[0057] If Pjk=0, the level division unit determines that the current monitoring period area is U jk The target heat level of the grid is level 1;

[0058] If 0<Pjk≤p1, the level division unit determines that the current monitoring period area is U jk The target heat level of the grid is level 2;

[0059] If Pjk>p1, the level division unit determines that the current monitoring period area is U jk The target heat level of the grid is three levels, where p1 is the preset target threshold, and Pjk is the number of valid targets divided into the grid with area Ujk during the monitoring period; by combining the number of monitored targets with the preset target threshold, the degree of insect pest in a certain grid can be expressed in a result-oriented manner, thereby improving the accuracy of the division of each grid type.

[0060] It can be understood that the present embodiment does not specifically limit the setting of the preset target threshold, and those skilled in the art can freely set it as long as the setting requirements of the preset target threshold are met, wherein the best value of p1 is 20.

[0061] Please continue reading Figure 3 As shown, the map generation module also includes a region division unit, which is connected to the level division unit, and the region division unit is used to divide each grid type according to the division result of the target heat level of each grid in each monitoring period within the management period to reflect the pest change trend, wherein:

[0062] If (w1×sjk1+w2×sjk2) / S=0, the area division unit determines that the current management period area is Ujk The grid type of the grid is safe area;

[0063] If (w1×sjk1+w2×sjk2) / S≤T, the area division unit determines that the current management period area is U jk The grid type of the grid is medium-risk area;

[0064] If (w1×sjk1+w2×sjk2) / S>T, the area division unit determines that the current management period area is U jk The grid type of the grid is a high-risk area; by identifying different types of areas, it provides an accurate basis for prevention and control work, ensures the reasonable distribution of control measures, improves safety, and improves the accuracy of disinfection map generation;

[0065] Among them, w1 is the weight of the grid with target heat level of level 2, w2 is the weight of the grid with target heat level of level 3, w1<w2, sjk1 is the weight of the grid with area U in the management period jk The number of monitoring cycles with grid target heat level of level 2, sjk2 is the area U within the management period jk The number of monitoring cycles with grid target heat level 3, S is the number of monitoring cycles in the management period, and T is the preset proportion threshold.

[0066] It can be understood that the preset proportion threshold and the setting of each weight are not specifically limited in this embodiment, and those skilled in the art can set them freely as long as they meet the setting requirements of the preset proportion threshold and each weight, among which the optimal value of T is 0.1, w1 is 0.8, and the optimal value of w2 is 1.2.

[0067] Please continue reading Figure 3 As shown, the map generation module also includes a map generation unit, which is connected to the area division unit. The map generation unit is used to generate a disinfection map according to the division results of each grid type within the management period, and output the disinfection map to the user.

[0068] Specifically, in this embodiment, the generation method of the disinfection map is not specifically limited. Technical personnel in this field can set it freely as long as the generation requirements of the disinfection map are met. Among them, the grid type of the safe area can be set to green, the grid type of the medium-risk area can be set to yellow, and the grid type of the high-risk area can be set to red. The disinfection map can be visualized through the Matpl ot lib library to generate a disinfection map.

[0069] Please continue reading Figure 1As shown, the system also includes a confidence analysis module, which is connected to the map generation module, and the confidence analysis module is used to adjust the generation process of the disinfection map according to the confidence of the effective targets of each grid within the management period.

[0070] See also Figure 4 As shown, the confidence analysis module includes:

[0071] The confidence analysis unit is used to analyze the abnormality of the confidence according to the confidence of the effective target of each grid within the management period, and adjust the division process of each grid type according to the analysis result, so as to dynamically adjust the system according to the average confidence of the effective target, wherein:

[0072] The confidence analysis unit manages the area within the period as U jk The average confidence of the valid targets in the grid is set to Zjk, and set: zr is the area U jk The confidence of the rth valid target in the grid, R is the area U jk The number of valid targets in the grid;

[0073] The confidence analysis unit defines the area as U jk The average confidence Zjk of the valid targets in the grid is compared with the preset threshold ZX0, and the abnormality of the confidence is analyzed according to the comparison result, and the area U is analyzed according to the comparison result. jk The mesh type division process of the mesh is adjusted. If Zjk≤ZX0, the confidence analysis unit determines that the area is U jk The confidence analysis unit determines that the area is U. jk The confidence of the effective targets in the grid is normal and no adjustment is made. By analyzing the confidence of the effective targets, it is possible to identify which grids have unreliable data, thereby improving the overall data quality. The system is dynamically adjusted according to the average confidence of the effective targets, so that the model can maintain efficient and accurate detection capabilities in different environments.

[0074] It can be understood that the present embodiment does not specifically limit the setting of the preset threshold, and those skilled in the art can freely set it as long as the setting requirements of the preset threshold are met, wherein the best value of ZX0 is 0.6.

[0075] Please continue reading Figure 4 As shown, the confidence analysis module also includes:

[0076] A construction unit is connected to the confidence analysis unit, and the construction unit is used to determine the confidence level according to the area U in the management period. jk The confidence of the effective target in the grid is used to construct the fluctuation coefficient of the confidence. The fluctuation analysis unit takes the area U within the management period as jk The fluctuation coefficient of the confidence of the valid target in the grid is set to BA, and set:

[0077]

[0078] Please continue reading Figure 4 As shown, the confidence analysis module also includes:

[0079] A fluctuation analysis unit is connected to the construction unit, and the fluctuation analysis unit manages the area U in the cycle jk The fluctuation coefficient BA of the confidence of the valid target in the grid is compared with the preset fluctuation coefficient B0, and the volatility of the confidence is analyzed according to the comparison result, and the analysis process of the abnormality of the confidence is adjusted according to the analysis result of the volatility. If BA≤B0, the fluctuation analysis unit determines that the area in the current management period is U jk The volatility of the confidence of the valid target in the grid is normal fluctuation and no adjustment is made. If BA>B0, the fluctuation analysis unit determines that the area in the current management period is U jk The volatility of the confidence of the valid targets in the grid is considered abnormal fluctuation, and the preset threshold is adjusted to ZX1, setting ZX1=ZX0×exp(B0-BA); by quantitatively analyzing the volatility, the accuracy of the disinfection map generation can be improved. At the same time, managers can take more scientific measures, make reasonable estimates of the disinfection effect, and enhance the scientific nature of disinfection decisions.

[0080] It can be understood that the present embodiment does not specifically limit the setting of the preset fluctuation coefficient, and those skilled in the art can freely set it as long as the setting requirements of the preset fluctuation coefficient are met, wherein the optimal value of B0 is 0.15.

[0081] Specifically, the system described in this embodiment is applied to the generation of indoor pest control maps. Through clear parameter settings and efficient module design, an efficient closed loop is formed between image acquisition, target detection, data analysis and pest control map generation, thereby significantly improving the accuracy and efficiency of cockroach identification and pest control. At the same time, the scalability and flexibility of the system provide guarantees for adaptation to different environments.

[0082] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not limitations on the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the embodiments here. All obvious changes or modifications derived from the technical solution of the present invention are still within the protection scope of the present invention.

Claims

1. A computer vision-based pest control map generation system, characterized in that: include: The image acquisition module is used to collect the video stream of the camera and perform frame processing on the video stream to extract the image; The target recognition module is used to perform target detection on the extracted image according to the target detection model to obtain the confidence of each target, and is also used to analyze the validity of the target according to the confidence of each target to obtain the bounding box of the valid target; A mapping analysis module is used to construct the actual ground coordinates of the valid target according to the border of the valid target, the camera installation height and the camera focal length; A grid division module is used to grid the ground and divide the target according to the construction results of the actual ground coordinates of the valid target; A map generation module is used to divide the target heat level of each grid according to the target division results in the monitoring period, and is also used to generate a disinfection map according to the division results of the target heat level of each grid in the management period, and output the generated disinfection map to the user; The confidence analysis module is used to adjust the generation process of the disinfection map according to the confidence of the effective targets of each grid within the management cycle.

2. The computer vision-based elimination map generation system according to claim 1, characterized in that: The target recognition module is used to perform target detection on the extracted image according to the target detection model to obtain the confidence Zi of the i-th target in the image, and compare the confidence Zi of the i-th target in the image with the preset confidence threshold z0 to analyze the validity of the target. The validity of the target includes valid targets and invalid targets. If the i-th target is a valid target, it is used as the i-valid target, and the border of the i-valid target is set to Ki(xi_center,yi_center,wi,hi), xi_center is the horizontal coordinate of the center point of the i-valid target border, yi_center is the vertical coordinate of the center point of the i-valid target border, wi is the width of the i-valid target bounding box, and hi is the height of the i-valid target bounding box.

3. The computer vision-based elimination map generation system according to claim 2, characterized in that: The mapping analysis module is used to construct the actual ground coordinates of the effective target according to the frame of the effective target, the camera installation height c and the camera focal length f, so as to obtain the actual ground coordinates (Sxi, Syi) of the center point of the i effective target.

4. The computer vision-based elimination map generation system according to claim 3 is characterized in that: The grid division module includes a grid division unit, which is used to divide the long side of the circumscribed rectangle of the ground into n equal parts, and divide the short side of the circumscribed rectangle of the ground into m equal parts, so as to divide the circumscribed rectangle of the ground into m×n grids.

5. The computer vision-based elimination map generation system according to claim 4, characterized in that: The grid division module further includes a target division unit, which is used to divide the target according to the grid division result and the construction result of the actual ground coordinates of the valid target, wherein: If (Sxi,Syi)∈U jk {x, y|(j-1)×L1 / n≤x<j×L1 / n, (k-1)×L2 / m≤y<k×L2 / m}, the target partitioning unit partitions the i valid targets into U jk Otherwise, the target partitioning unit does not partition i valid targets into U jk , where L1 is the length of the long side of the circumscribed rectangle of the ground, L2 is the length of the short side of the circumscribed rectangle of the ground, 0<j≤n, 0<k≤m, U jk It is the area of ​​the grid whose long side is j and short side is k in the circumscribed rectangle of the ground.

6. The computer vision-based elimination map generation system according to claim 5, characterized in that: The map generation module includes a level division unit, which is used to divide the target heat level of each grid according to the target division result within the monitoring period, and the target heat level includes level one, level two and level three.

7. The computer vision-based elimination map generation system according to claim 6, characterized in that: The map generation module further includes a region division unit, which is used to divide each grid type according to the division result of the target heat level of each grid in each monitoring period within the management period, wherein: If (w1×sjk1+w2×sjk2) / S=0, the area division unit determines that the grid type of the grid whose current management period area is Ujk is a safe area; If (w1×sjk1+w2×sjk2) / S≤T, the area division unit determines that the grid type of the grid whose current management period area is Ujk is a medium-risk area; If (w1×sjk1+w2×sjk2) / S>T, the area division unit determines that the grid type of the grid whose current management period area is Ujk is a high-risk area; Among them, w1 is the weight of the grid with target heat level of level 2, w2 is the weight of the grid with target heat level of level 3, sjk1 is the weight of the grid with area U in the management period jk The number of monitoring cycles with grid target heat level of level 2, sjk2 is the area U within the management period jk The number of monitoring cycles with grid target heat level 3, S is the number of monitoring cycles in the management period, and T is the preset proportion threshold.

8. The computer vision-based elimination map generation system according to claim 7, characterized in that: The confidence analysis module includes a confidence analysis unit, which is used to analyze the abnormality of confidence according to the confidence of each grid valid target within the management period. The abnormality of confidence includes normal and abnormal. If the area is U jk The confidence level of the grid valid target is abnormal, and the preset abnormal proportion threshold is set to T1.

9. The computer vision-based elimination map generation system according to claim 8, characterized in that: The confidence analysis module also includes a construction unit, which is used to calculate the confidence level of the area according to the management period. jk The confidence of the effective target in the grid is used to construct the fluctuation coefficient of the confidence to obtain the area U within the management period. jk The fluctuation coefficient BA of the confidence of valid targets in the grid.

10. The computer vision-based elimination map generation system according to claim 9, characterized in that: The confidence analysis module also includes a fluctuation analysis unit, which manages the area U within the period. jk The fluctuation coefficient BA of the confidence of the effective target in the grid is compared with the preset fluctuation coefficient B0, and the volatility of the confidence is analyzed according to the comparison result. The volatility of the confidence includes normal fluctuation and abnormal fluctuation. If the area in the current management cycle is U jk The fluctuation of the confidence of valid targets in the grid is abnormal fluctuation, and the preset threshold is adjusted to ZX1.

Images