Patrol robot image recognition method and computer storage medium
Patent Information
- Application Number
- CN202411134371.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2024-12-03
Smart Images

Figure CN119068432A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of patrol robot image recognition, in particular to a patrol robot image recognition method and a computer storage medium. Background Art
[0002] At present, patrol robots are often used in various areas such as industrial parks, smart buildings, and parking lots. However, in the parking lot scenario, the scope of use is still relatively small. In the existing technology, most patrol robots used in parking lots are only used to patrol and take pictures along fixed routes, which is difficult to adapt to today's intelligent trend and is even more unfavorable for intelligent garage management. Therefore, the patrol robot with image recognition technology is combined with the existing garage management technology to realize efficient and intelligent garage management. There is still a lot of room for improvement in this application.
[0003] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention
[0004] The object of the present invention is to provide a patrol robot image recognition method and a computer storage medium to solve the problems raised in the above background technology.
[0005] To achieve the above object, the present invention provides the following technical solutions:
[0006] A patrol robot image recognition method, the specific steps include:
[0007] S1: Collect owner information, add tag data to the preset parking space information based on the owner information to form a database, then collect license plate information of vehicles entering the parking lot, compare the license plate information with the database, divide the vehicles entering the parking lot into owner vehicles and outside vehicles, collect the entry time of owner vehicles and outside vehicles respectively, and execute step S2 for owner vehicles and execute step S3 for outside vehicles;
[0008] S2: Collect the exit time of the owner's vehicle, set a time observation window to perform time series analysis on the entry and exit time of the owner's vehicle, generate parking intervals for the owner's vehicle on weekdays and holidays, generate redundant intervals based on the parking intervals and owner information, and use the redundant intervals to update the label data;
[0009] S3: Collect the expected parking time of external vehicles, generate a recommendation index for each parking space based on the entry time, expected parking time, redundant interval, parking space information, and owner information of external vehicles, and then use patrol robots to guide external vehicles to the parking space with the highest recommendation index along a preset movement route. At the same time, collect the video stream of external vehicles during the parking process, extract the image frames of the video stream and process them using the pre-trained YOLOv5 model to determine the parking status of the vehicle;
[0010] S4: The patrol robot patrols regularly according to the expected parking time of the foreign vehicle, collects the parking images of the foreign vehicle for storage, and calculates the remaining time according to the expected parking time and the actual parking time of the foreign vehicle. When the remaining time is less than the preset time threshold, an alarm is issued to the foreign vehicle.
[0011] Preferably, the owner information includes the owner's license plate information, parking space number, and authorization information, and the parking space information includes a garage map and parking space coordinates. The owner's license plate information is bound to the parking space number, authorization information, parking space coordinates, redundant interval, and redundant interval to form tag data, and the tag data is used to update the garage map according to the parking space number, thereby generating a database based on the garage map, wherein the owner's authorization information is a dynamic data Q allow , when the owner allows the temporary parking of external vehicles, dynamic data Q allow =1, when the owner does not allow temporary parking of external vehicles, dynamic data Q allow =0.
[0012] Preferably, the logic for generating redundant intervals is:
[0013] The time observation window is divided into the first observation window and the second observation window according to working days and holidays. The entry and exit times of the owner's vehicles in the first observation window and the second observation window are counted respectively. The parking time distribution diagram of the owner's vehicles is generated with time as the horizontal axis and the number of parking times as the vertical axis. The horizontal axis is evenly divided into M intervals A j , each interval A j The time length is t, t = 24 / M, and the subscript represents the interval number, j = 1, 2, 3, ..., M;
[0014] The daily parking time distribution graphs of the owner's vehicles in the first observation window and the second observation window are respectively superimposed, and the parking times of the ordinates of the overlapping parts of the graphs are increased by one to obtain the total parking time distribution graph of the owner's vehicles on weekdays and holidays;
[0015] Calculate the corresponding redundant interval based on the total parking time distribution of the owner's vehicle on weekdays and holidays in All represent the interval with ordinate 0 in the total parking time distribution diagram, and the subscript j 1 Indicates the interval number with ordinate 0 in the total parking time distribution diagram, redundant interval The subscript i represents the parking space number, and the superscript g represents the distinction between weekdays and holidays. When g=1, it is the redundant interval of weekdays, and when g=0, it is the redundant interval of holidays.
[0016] Preferably, whether it is a working day is determined according to the entry time of the foreign vehicle, and the temporary parking interval Lt is generated according to the entry time of the foreign vehicle and the expected parking time g , and then calculate the recommended index of parking spaces based on the temporary parking interval and redundant interval The calculation method is:
[0017]
[0018] Where L i x, L i y represents the horizontal and vertical coordinates of the parking space numbered i in the garage map, L r x, L r y represents the horizontal and vertical coordinates of the garage entrance in the garage map, E allow Indicates the time judgment index.
[0019] Preferably, the logic for determining the parking status of the vehicle is:
[0020] Collect the video stream of the external vehicle during the parking process and extract the image frames of the video stream;
[0021] Input the image frames into the pre-trained Y0LOv5 model and output the bounding box of the vehicle and the parking space in each image frame;
[0022] The vehicle boundary box corresponding to the image frame during the parking process is compared with the parking space boundary box to determine whether the foreign vehicle is parked in a standardized manner during the parking process. The vehicle boundary box corresponding to the image frame after parking is completed is compared with the parking space boundary box to determine whether the foreign vehicle is parked in a standardized manner after parking is completed.
[0023] Preferably, the number of image frames extracted from the video stream is calibrated as P, and the first P-1 image frames are divided into a process group, and the last image frame is divided into a result group.
[0024] Preferably, the coordinates of the four vertices of the vehicle bounding box in each image frame in the process group are calibrated as The coordinates of the four vertices of the parking space bounding box are calibrated as The subscript p represents the number of the image frame, p=1, 2, 3, ..., P-1, and the superscript d represents the vertex number, d=1, 2, 3, 4. The area of the vehicle bounding box in the result group image frame is calibrated as Sc, and the area of the parking space bounding box is calibrated as Sw.
[0025] Preferably, if there are image frames in the process group that satisfy:
[0026]
[0027] If the parking state of the foreign vehicle is considered to be crossing the line, the patrol robot will sound an alarm;
[0028] If the image frames of the result group satisfy:
[0029] Sc∪Sw>Sw
[0030] If the parking state of the foreign vehicle is considered to be crossing the line, the patrol robot will sound an alarm;
[0031] in Respectively represent the minimum and maximum values of the horizontal coordinates of the four vertices of the vehicle bounding box, They respectively represent the minimum and maximum values of the horizontal coordinates of the four vertices of the parking space bounding box.
[0032] A computer storage medium is used to store a computer program, and the computer program implements the above-mentioned identification method when executed by a processor.
[0033] Compared with the prior art, the present invention has the following beneficial effects:
[0034] The present invention combines image recognition technology with the existing garage management system to achieve efficient and accurate vehicle identification and parking status monitoring. Compared with the traditional garage management method, the use of the YOLOv5 model for image recognition not only improves the accuracy and real-time performance of detection, but also automatically identifies and handles illegal parking behaviors, reducing manual intervention. At the same time, combined with the owner's authorization information and dynamic data analysis, the solution achieves the optimization and flexible management of parking space utilization. The combination of these technologies not only improves the intelligent level of garage management, ensures the standardization and safety of the garage, but also greatly improves the user's parking experience, reflects the significant advantages of intelligent garage management, and at the same time increases the economic benefits of the owner, further expanding the market competitiveness and applicability of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 It is a schematic diagram of the overall method flow of the present invention. DETAILED DESCRIPTION
[0036] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.
[0037] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.
[0038] Example:
[0039] See also Figure 1 , the present invention provides a technical solution:
[0040] A patrol robot image recognition method, the specific steps include:
[0041] S1: Collect owner information, add label data to the preset parking space information based on the owner information to form a database, then collect license plate information of the entering vehicles, compare the license plate information with the database, divide the entering vehicles into owner vehicles and outside vehicles, collect the entry time of owner vehicles and outside vehicles respectively, and execute step S2 for owner vehicles and step S3 for outside vehicles.
[0042] The owner information includes the owner's license plate information, parking space number, and authorization information. The parking space information includes a garage map and parking space coordinates. The owner's license plate information is bound to the parking space number, authorization information, parking space coordinates, redundant intervals, and redundant intervals to form label data. The label data is used to update the garage map according to the parking space number, thereby generating a database based on the garage map, in which the owner's authorization information is a dynamic data Q allow , when the owner allows the temporary parking of external vehicles, dynamic data Q allow =1, when the owner does not allow temporary parking of external vehicles, dynamic data Q allow =0.
[0043] In this step, by integrating the owner information and parking space information, the database structure becomes clearer and more organized, which is convenient for subsequent query and management. At the same time, the unified management of label data can quickly and accurately identify the vehicle type (owner's vehicle or external vehicle), improve the system's response speed and accuracy, and by setting the owner's authorization information, flexible adjustment of parking space usage rights can be achieved, significantly improving parking space utilization.
[0044] S2: Collect the exit time of the owner's vehicle, set a time observation window to perform time series analysis on the entry and exit time of the owner's vehicle, generate parking intervals for the owner's vehicle on weekdays and holidays respectively, generate redundant intervals based on the parking intervals and owner information, and use the redundant intervals to update the label data.
[0045] The logic for generating redundant intervals is:
[0046] The time observation window is divided into the first observation window and the second observation window according to working days and holidays. The entry and exit times of the owner's vehicles in the first observation window and the second observation window are counted respectively. The parking time distribution diagram of the owner's vehicles is generated with time as the horizontal axis and the number of parking times as the vertical axis. The horizontal axis is evenly divided into M intervals A j , each interval A j The time length is t, t = 24 / M, and the subscript represents the interval number, j = 1, 2, 3, ..., M;
[0047] The daily parking time distribution graphs of the owner's vehicles in the first observation window and the second observation window are respectively superimposed, and the parking times of the ordinates of the overlapping parts of the graphs are increased by one to obtain the total parking time distribution graph of the owner's vehicles on weekdays and holidays;
[0048] Calculate the corresponding redundant interval based on the total parking time distribution of the owner's vehicle on weekdays and holidays in All represent the interval with ordinate 0 in the total parking time distribution diagram, and the subscript j 1 Indicates the interval number with ordinate 0 in the total parking time distribution diagram, redundant interval The subscript i represents the parking space number, and the superscript g represents the distinction between weekdays and holidays. When g=1, it is the redundant interval of weekdays, and when g=0, it is the redundant interval of holidays.
[0049] Here is an example to illustrate. Suppose there are two households, Owner 1 and Owner 2, and the preset time length t = 0.5h, then the corresponding M = 48, that is, from 00:00 of each day to 00:00 of the next day, every half hour is an interval, and there are A 1 ~~A48 interval, A 1 =[00:00,00:30], A 2 =[00:30,01:00],…,A 48 =[11:30,00:00], the time observation window is 30 days, including 22 days for the first observation window and 8 days for the second observation window.
[0050] In the initial state, their redundant intervals have no value and can be represented by null in the label data, that is, At this time, the data contained in the corresponding parking space will not be included in the subsequent calculation, that is, it will not be recommended to external vehicles. After 30 days of observation, the entry time of owner 1 on weekdays is distributed between 17:00 and 17:20, and the exit time is distributed between 08:10 and 08:20. Correspondingly, the A of that day 17 ~A 34 The number of parking times in these intervals is 0, and the vertical coordinate in the parking time distribution diagram is 0, which is the redundant interval. It reflects the time period when the owner does not park for a long time. Similarly, if the exit and entry times of owner 2 in the morning and evening are the same as those of owner 1, but he is used to going home for a nap at noon on weekdays, and his entry time at noon is distributed in the range of 12:00-12:20, and his exit time is distributed in the range of 13:50-14:00, then the redundant interval of owner 2 on weekdays
[0051] The longer the redundant interval is, the longer the corresponding parking space is vacant, and the more suitable it is for parking external vehicles. The calculation method for holidays is the same, so I will not elaborate on it here.
[0052] In this step, the time observation window is divided into two parts: weekdays and holidays, and the parking time distribution of the owner's vehicle is counted separately. This can more accurately analyze the parking behavior of the owner's vehicle, including the different parking patterns on weekdays and holidays, which helps to more accurately predict and plan the use of parking spaces, improve parking space utilization, and enable managers to manage and dispatch garages more efficiently, reduce manual intervention, and improve management efficiency.
[0053] S3: Collect the expected parking time of external vehicles, generate a recommendation index for each parking space based on the entry time, expected parking time, redundant interval, parking space information, and owner information of external vehicles, and then use the patrol robot to guide external vehicles to the parking space with the highest recommendation index using a preset movement route. At the same time, collect the video stream of external vehicles during the parking process, extract the image frames of the video stream and process them using the pre-trained YOL0v5 model to determine the parking status of the vehicle.
[0054] Determine whether it is a working day based on the entry time of the external vehicle, and generate the temporary parking interval Lt based on the entry time and expected parking time of the external vehicle g , and then calculate the recommended index of parking spaces based on the temporary parking interval and redundant interval The calculation method is:
[0055]
[0056]
[0057] Where L i x, L i y represents the horizontal and vertical coordinates of the parking space numbered i in the garage map, L r x, L r y represents the horizontal and vertical coordinates of the garage entrance in the garage map, E allow Indicates the time judgment index.
[0058] From the calculation model of the recommendation index, it can be seen that the recommendation index is mainly divided into three parts. The first part is the owner's authorization information Q allow , used to control the overall recommendation index. When the owner authorizes, the recommendation index is calculated normally. When the owner does not authorize, the recommendation index is 0, and external vehicles are not allowed to park, avoiding disputes between external vehicles and owners. The second part is the time judgment index E allow , and its value is obtained as follows:
[0059]
[0060] This part indicates the overlap between the temporary parking interval of external vehicles and the redundant interval of parking spaces. If the temporary parking interval of external vehicles exceeds the redundant interval of parking spaces, then This means that it may affect the normal entry and exit of the owner, so it is directly set to 0, and external vehicles are not allowed to park. On the contrary, if the temporary parking area of external vehicles is within the redundant area of the parking space, then it will meet This means that the normal access of the owners will not be affected, so 1 is directly selected to allow external vehicles to park;
[0061] The final third part is:
[0062]
[0063] This part reflects the distance between the parking space and the garage entrance. When the owner's authorization information and time judgment index are met, the distance between the parking space and the garage entrance is inversely proportional to the recommendation index. That is to say, parking spaces close to the garage entrance are recommended to foreign vehicles to prevent them from getting lost in unfamiliar garages and affecting the normal entry and exit of owners.
[0064] In this step, the owner's authorization information can control the use of parking spaces in real time, avoid disputes between external vehicles and owners, flexibly manage temporary use rights of parking spaces, improve parking space utilization, and at the same time ensure that the rights and interests of owners are not infringed. The time determination index ensures that the temporary parking area of external vehicles will not affect the normal entry and exit of the owner, and ensures that the parking behavior of external vehicles will not interfere with the owner, thereby improving the reliability of the system and user satisfaction. Through flexible parking space management and intelligent recommendations, the utilization rate of parking spaces is improved and the waste of parking space resources is avoided.
[0065] S4: The patrol robot patrols regularly according to the expected parking time of the foreign vehicle, collects the parking images of the foreign vehicle for storage, and calculates the remaining time according to the expected parking time and the actual parking time of the foreign vehicle. When the remaining time is less than the preset time threshold, an alarm is issued to the foreign vehicle.
[0066] The logic for determining the parking status of a vehicle is:
[0067] Collect the video stream of the external vehicle during the parking process and extract the image frames of the video stream;
[0068] Input the image frames into the pre-trained Y0LOv5 model and output the bounding box of the vehicle and the parking space in each image frame;
[0069] The vehicle boundary box corresponding to the image frame during the parking process is compared with the parking space boundary box to determine whether the foreign vehicle is parked in a standardized manner during the parking process. The vehicle boundary box corresponding to the image frame after parking is completed is compared with the parking space boundary box to determine whether the foreign vehicle is parked in a standardized manner after parking is completed.
[0070] In this step, the patrol robot can collect video streams through the camera, and at the same time, the camera can be set on a lifting rod that can rise and fall, and shoot from a bird's-eye view, so as to better reflect the positional relationship between the vehicle bounding box and the parking space bounding box. The extraction of image frames can be implemented using the OpenCV image library and Python. The Y0L0v5 model can collect a large number of parking lot images in advance. These images should include various parking states (such as standardized parking, line parking, empty parking spaces, etc.), and use tools (such as Labelmg) to annotate the images. The annotation content includes the vehicle's bounding box and the parking space's bounding box. The annotation results are usually saved in YOLO format, including the image name, bounding box category, bounding box center coordinates and width and height, and then input into the YOLOv5 model for training. Since the extraction of image frames and the YOLOv5 model are both existing technologies, their specific principles will not be repeated.
[0071] The number of image frames extracted from the video stream is calibrated as P, and the first P-1 image frames are divided into the process group, and the last image frame is divided into the result group. The coordinates of the four vertices of the vehicle bounding box in each image frame in the process group are calibrated as The coordinates of the four vertices of the parking space bounding box are calibrated as The subscript p represents the number of the image frame, p=1, 2, 3, ..., P-1, and the superscript d represents the vertex number, d=1, 2, 3, 4. The area of the vehicle bounding box in the result group image frame is calibrated as Sc, and the area of the parking space bounding box is calibrated as Sw.
[0072] If there are image frames in the process group that satisfy:
[0073]
[0074] It is considered that the parking state of the foreign vehicle during the parking process is crossing the line, and the patrol robot issues an alarm.
[0075] in Respectively represent the minimum and maximum values of the horizontal coordinates of the four vertices of the vehicle bounding box, They respectively represent the minimum and maximum values of the horizontal coordinates of the four vertices of the parking space bounding box.
[0076] The judgment logic here is: it is normal for one side of an external vehicle to exceed the parking space boundary box when reversing into a parking space. If the above conditions are met, it proves that both sides of the external vehicle have exceeded the parking space boundary box. It is considered that the reversing angle is incorrect, causing the vehicle to be too tilted and there is a risk of collision, so an alarm is issued.
[0077] If the image frames of the result group satisfy:
[0078] Sc∪Sw>Sw
[0079] The patrol robot will then determine that the parking state of the foreign vehicle after parking is over the line, and issue an alarm.
[0080] The judgment logic here is: when an external vehicle reverses into the parking space normally, its vehicle boundary box should be located within the parking space boundary box, that is, Sc∪Sw=Sw. If Sc∪Sw>Sw, it is considered that the external vehicle has crossed the line or exceeded the line, so an alarm is issued.
[0081] In this step, patrol robots are used to guide and detect incoming vehicles. Patrol robots can automatically patrol and collect images according to predetermined time intervals to achieve all-weather automated parking management. Combined with the real-time detection capabilities of the YOLOv5 model, parking violations can be discovered and handled in a timely manner to ensure that vehicles remain in compliance throughout the parking period. It can also reduce manual intervention and improve management efficiency.
[0082] A computer storage medium is used to store a computer program, and the computer program implements the above-mentioned identification method when executed by a processor.
[0083] In summary, the present invention combines image recognition technology with the existing garage management system to achieve efficient and accurate vehicle identification and parking status monitoring. Compared with the traditional garage management method, the use of the Y0L0v5 model for image recognition not only improves the accuracy and real-time performance of detection, but also automatically identifies and processes illegal parking behaviors, reducing manual intervention. At the same time, combined with the owner's authorization information and dynamic data analysis, the solution achieves the optimization and flexible management of parking space utilization. The combination of these technologies not only improves the intelligent level of garage management, ensures the standardization and safety of the garage, but also greatly improves the user's parking experience, reflects the significant advantages of intelligent garage management, and also improves the economic benefits of the owner, further expanding the market competitiveness and applicability of the present invention.
[0084] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.
[0085] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.
[0086] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0087] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.
Claims
1. A patrol robot image recognition method, characterized in that: The specific steps include: S1: Collect owner information, add tag data to the preset parking space information based on the owner information to form a database, then collect license plate information of vehicles entering the parking lot, compare the license plate information with the database, divide the vehicles entering the parking lot into owner vehicles and outside vehicles, collect the entry time of owner vehicles and outside vehicles respectively, and execute step S2 for owner vehicles and execute step S3 for outside vehicles; S2: Collect the exit time of the owner's vehicle, set a time observation window to perform time series analysis on the entry and exit time of the owner's vehicle, generate parking intervals for the owner's vehicle on weekdays and holidays, generate redundant intervals based on the parking intervals and owner information, and use the redundant intervals to update the label data; S3: Collect the expected parking time of external vehicles, generate a recommendation index for each parking space based on the entry time, expected parking time, redundant interval, parking space information, and owner information of external vehicles, and then use patrol robots to guide external vehicles to the parking space with the highest recommendation index along a preset movement route. At the same time, collect the video stream of external vehicles during the parking process, extract the image frames of the video stream and process them using the pre-trained YOLOv5 model to determine the parking status of the vehicle; S4: The patrol robot patrols regularly according to the expected parking time of the foreign vehicle, collects the parking images of the foreign vehicle for storage, and calculates the remaining time according to the expected parking time and the actual parking time of the foreign vehicle. When the remaining time is less than the preset time threshold, an alarm is issued to the foreign vehicle.
2. The patrol robot image recognition method according to claim 1, characterized in that: The owner information includes the owner's license plate information, parking space number, and authorization information. The parking space information includes a garage map and parking space coordinates. The owner's license plate information is bound to the parking space number, authorization information, parking space coordinates, redundant intervals, and redundant intervals to form label data. The label data is used to update the garage map according to the parking space number, thereby generating a database based on the garage map, in which the owner's authorization information is a dynamic data Q allow , when the owner allows the temporary parking of external vehicles, dynamic data Q allow =1, when the owner does not allow temporary parking of external vehicles, dynamic data Q allow =0.
3. A patrol robot image recognition method according to claim 2, characterized in that: The logic for generating redundant intervals is: The time observation window is divided into the first observation window and the second observation window according to working days and holidays. The entry and exit times of the owner's vehicles in the first observation window and the second observation window are counted respectively. The parking time distribution diagram of the owner's vehicles is generated with time as the horizontal axis and the number of parking times as the vertical axis. The horizontal axis is evenly divided into M intervals A j , each interval A j The time length is t, t = 24 / M, and the subscript j represents the interval number, j = 1, 2, 3, ..., M; The daily parking time distribution graphs of the owner's vehicles in the first observation window and the second observation window are respectively superimposed, and the parking times of the ordinates of the overlapping parts of the graphs are increased by one to obtain the total parking time distribution graph of the owner's vehicles on weekdays and holidays; Calculate the corresponding redundant interval based on the total parking time distribution of the owner's vehicle on weekdays and holidays in All of them represent the interval with ordinate 0 in the total parking time distribution diagram. The subscript j1 represents the interval number with ordinate 0 in the total parking time distribution diagram. The redundant interval The subscript i represents the parking space number, and the superscript g represents the distinction between weekdays and holidays. When g=1, it is the redundant interval of weekdays, and when g=0, it is the redundant interval of holidays.
4. The patrol robot image recognition method according to claim 3, characterized in that: Determine whether it is a working day based on the entry time of the external vehicle, and generate the temporary parking interval Lt based on the entry time and expected parking time of the external vehicle g , and then calculate the recommended index of parking spaces based on the temporary parking interval and redundant interval The calculation method is: Where L i x, L i y represents the horizontal and vertical coordinates of the parking space numbered i in the garage map, L r x, L r y represents the horizontal and vertical coordinates of the garage entrance in the garage map, E allow Indicates the time judgment index.
5. The patrol robot image recognition method according to claim 1, characterized in that: The logic for determining the parking status of a vehicle is: Collect the video stream of the external vehicle during the parking process and extract the image frames of the video stream; Input the image frames into the pre-trained Y0LOv5 model and output the bounding box of the vehicle and the parking space in each image frame; The vehicle boundary box corresponding to the image frame during the parking process is compared with the parking space boundary box to determine whether the foreign vehicle is parked in a standardized manner during the parking process. The vehicle boundary box corresponding to the image frame after parking is completed is compared with the parking space boundary box to determine whether the foreign vehicle is parked in a standardized manner after parking is completed.
6. A patrol robot image recognition method according to claim 5, characterized in that: The number of image frames extracted from the video stream is calibrated as P, and the first P-1 image frames are divided into a process group, and the last image frame is divided into a result group.
7. A patrol robot image recognition method according to claim 6, characterized in that: The coordinates of the four vertices of the vehicle bounding box in each image frame in the process group are calibrated as The coordinates of the four vertices of the parking space bounding box are calibrated as The subscript p represents the number of the image frame, p=1, 2, 3, ..., P-1, and the superscript d represents the vertex number, d=1, 2, 3, 4. The area of the vehicle bounding box in the result group image frame is calibrated as Sc, and the area of the parking space bounding box is calibrated as Sw.
8. The patrol robot image recognition method according to claim 7, characterized in that: If there are image frames in the process group that satisfy: If the parking state of the foreign vehicle is considered to be crossing the line, the patrol robot will sound an alarm; If the image frames of the result group satisfy: Sc∪Sw>Sw If the parking state of the foreign vehicle is considered to be crossing the line, the patrol robot will sound an alarm; in Respectively represent the minimum and maximum values of the horizontal coordinates of the four vertices of the vehicle bounding box, They respectively represent the minimum and maximum values of the horizontal coordinates of the four vertices of the parking space bounding box.
9. A computer storage medium, characterized in that: The storage medium is used to store a computer program, and when the computer program is executed by a processor, the identification method according to any one of claims 1 to 8 is implemented.
