Image recognition-based parking lot monitoring method, device, equipment and medium

Abstract

The application discloses a parking lot monitoring method and device based on image recognition, equipment and medium, the method comprises: based on image recognition of parking lot monitoring method and device, equipment and medium, the method comprises: according to the parking monitoring period from video stream intercepts video image and extracts image features, obtains the environment characteristics corresponding to video image, the environment characteristics and image features are combined and input to the abnormality identification model to obtain the monitoring identification result of video image, it is judged whether the monitoring identification result is abnormal, if there is no abnormality, continue to execute the step of intercepting video image, if there is abnormality, send alarm prompt information.The parking lot monitoring method can utilize the video stream of existing video monitor to monitor and analyze the intercepted video image, so as to not be limited by the display size of video picture, clear video image intercepted in the original video stream can be analyzed, and the efficiency and reliability of abnormal monitoring of vehicles and charging facilities in the parking lot are improved.

Description

Image recognition-based parking lot monitoring methods, devices, equipment, and media Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a parking lot monitoring method, device, equipment, and medium based on image recognition. Background Technology

[0002] New energy vehicles require battery charging, which can lead to abnormal situations such as battery combustion and charging facility overheating in parking lots. Traditional methods typically involve manual monitoring of parked new energy vehicles. However, the number of service personnel is limited, and monitoring can only be conducted at long intervals (e.g., half an hour or an hour), sometimes failing to detect problems in a timely manner. Parking lots are usually equipped with multiple video surveillance cameras. Monitoring personnel can monitor parked new energy vehicles and related charging facilities, summarize and report any problems found, and facilitate timely handling of abnormalities to minimize losses. However, due to the large number of video monitors and the small size of each monitor's display, it is difficult to accurately identify details in the video footage; furthermore, a limited number of monitoring personnel cannot simultaneously monitor numerous video feeds, resulting in the inability to detect abnormalities in a timely manner. Therefore, existing methods that rely on manually reviewing monitor video feeds cannot accurately monitor abnormalities in parking lots. Summary of the Invention

[0003] This invention provides a parking lot monitoring method, device, equipment, and medium based on image recognition, aiming to solve the problem that existing methods cannot accurately monitor parking lot anomalies by manually viewing video footage from monitors.

[0004] In a first aspect, embodiments of the present invention provide a parking lot monitoring method based on image recognition. This method is applied to a management server, which establishes a network connection with at least one video surveillance camera to achieve data transmission. The method includes:

[0005] According to the preset parking monitoring cycle, the corresponding video image is extracted from the video stream input by the video monitor;

[0006] The corresponding image features are extracted from each of the video images according to the preset image feature extraction model;

[0007] According to the preset environmental feature acquisition rules, the environmental features corresponding to the video monitor to which each video image belongs are obtained;

[0008] The image features and environmental features of each video image are combined and input into a preset anomaly recognition model to perform anomaly recognition on each video image and obtain the corresponding monitoring recognition result.

[0009] Determine whether there are any anomalies in the monitoring and recognition results of each video image;

[0010] If the monitoring and recognition results of each video image are normal, return to the step of extracting the corresponding video image from the video stream input by the video monitor according to the preset parking monitoring cycle;

[0011] If any of the video images shows an abnormality in the monitoring and identification results, a corresponding alarm message will be issued.

[0012] Secondly, embodiments of the present invention also provide a parking lot monitoring device based on image recognition, wherein the device is configured in a management server, the management server establishes a network connection with at least one video surveillance camera to realize data information transmission, the device applies the parking lot monitoring method based on image recognition as described in the first aspect above, and the device includes:

[0013] The video image capture unit is used to capture corresponding video images from the video stream input by the video monitor according to a preset parking monitoring cycle.

[0014] An image feature extraction unit is used to extract corresponding image features from each of the video images according to a preset image feature extraction model;

[0015] An environmental feature acquisition unit is used to acquire environmental features corresponding to the video surveillance camera to which each video image belongs, according to a preset environmental feature acquisition rule.

[0016] The monitoring and identification result acquisition unit is used to combine the image features and environmental features of each video image and input them into a preset anomaly identification model to perform anomaly identification on each video image and obtain the corresponding monitoring and identification result.

[0017] An anomaly detection unit is used to determine whether there are any anomalies in the monitoring and recognition results of each video image;

[0018] The execution unit is used to return to the step of extracting the corresponding video image from the video stream input by the video monitor according to the preset parking monitoring cycle if the monitoring and recognition results of each video image are normal.

[0019] The alarm notification unit is used to issue a corresponding alarm notification if the monitoring and identification result of any of the video images is abnormal.

[0020] Thirdly, embodiments of the present invention also provide a computer device, wherein the device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0021] Memory, used to store computer programs;

[0022] When a processor executes a program stored in memory, it implements the steps of the image recognition-based parking monitoring method as described in the first aspect.

[0023] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the parking lot monitoring method based on image recognition as described in the first aspect.

[0024] This invention provides a parking lot monitoring method, apparatus, device, and medium based on image recognition. The method includes: capturing video images from a video stream according to a parking monitoring cycle and extracting image features; obtaining environmental features corresponding to the video images; merging the environmental features and image features and inputting them into an anomaly recognition model to obtain the monitoring and recognition result of the video images; determining whether the monitoring and recognition result is abnormal; if there is no abnormality, continuing the step of capturing video images; if there is an abnormality, issuing an alarm message. The above-described parking lot monitoring method can utilize the video stream of existing video monitors to capture and analyze video images, thus not being limited by the display size of the video screen. It can analyze clearer video images captured from the original video stream, significantly improving the efficiency and reliability of anomaly monitoring of vehicles and charging facilities in parking lots. It can accurately monitor abnormal problems occurring in parking lots using the video stream of video monitors. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 is a flowchart illustrating the parking lot monitoring method based on image recognition provided in an embodiment of the present invention;

[0027] Figure 2 is a schematic diagram of the application scenario of the parking lot monitoring method based on image recognition provided in the embodiment of the present invention;

[0028] Figure 3 is a schematic block diagram of a parking lot monitoring device based on image recognition provided in an embodiment of the present invention;

[0029] Figure 4 is a schematic block diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0032] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0033] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0034] Please refer to Figures 1 and 2. As shown in the figures, an embodiment of this invention provides a parking lot monitoring method based on image recognition. This method is applied to a management server 10. As shown in Figure 2, the management server 10 establishes network connections with at least one video monitor 20 to transmit data information. The management server 10 is a server located within the parking lot used to acquire video streams collected by each video monitor 20 and perform anomaly monitoring. The management server 10 can be a terminal device with video image processing and information display capabilities, such as a desktop computer or a monitoring management terminal. The video monitor 20 is a surveillance camera installed inside the parking lot, at the parking lot entrance, etc. The video monitor 20 can acquire real-time monitoring images to obtain video streams and transmit them to the management server 10. As shown in Figure 1, the method includes steps S110 to S170.

[0035] S110. Extract the corresponding video image from the video stream input by the video monitor according to the preset parking monitoring cycle.

[0036] According to a preset parking monitoring cycle, the video monitor extracts corresponding video images from the video stream input by the video monitor. The video monitor transmits the acquired video images to the management server in real time. The management server receives a set of video streams, with each set of video streams corresponding to one video monitor. The management server extracts corresponding video images from the video streams according to the preset parking monitoring cycle, so that one video image can be extracted from each set of video streams in one monitoring session. For example, the parking monitoring cycle can be set to 1 minute, then one video image can be extracted from each video stream every 1 minute.

[0037] S120. Extract corresponding image features from each of the video images according to the preset image feature extraction model.

[0038] Image features are extracted from each video image according to a pre-set image feature extraction model. Furthermore, image features can be extracted from each video image using the image feature extraction model, and these image features can be used to characterize the features of the video images. Specifically, the image feature extraction model consists of convolutional layers and pooling layers, which are alternately configured. The convolutional layers upsample the pixel values ​​of pixels in the video image to obtain lower-resolution data values; one upsampling process reduces both the number of horizontal and vertical pixels. The pooling layers reduce the dimensionality of the upsampled data values, i.e., perform dimensionality reduction; one dimensionality reduction process keeps the number of horizontal pixels unchanged while reducing the number of vertical pixels. For example, by extracting features from a 1920×1080 resolution video image using the image feature extraction model, a set of 1×256 arrays is obtained as image features; then, the image features corresponding to each video image can be obtained through this image feature extraction rule.

[0039] S130. Obtain the environmental features corresponding to the video monitor to which each video image belongs, according to the preset environmental feature acquisition rules.

[0040] According to preset environmental feature acquisition rules, environmental features corresponding to the video monitors to which each video image belongs can be obtained. Furthermore, environmental features corresponding to each video image can be obtained according to the environmental feature acquisition rules. These environmental features are used to characterize the environment in which each video image is located. During this process, the video monitor to which the video image belongs can be determined, and the environmental features corresponding to each video monitor can be obtained according to the environmental feature acquisition rules.

[0041] In a specific embodiment, step S130 specifically includes the following sub-steps: determining the corresponding location type feature based on the installation location of the video surveillance device to which the video image belongs; obtaining matching illumination features from the environmental feature acquisition rules based on the current time point and the installation location corresponding to the video image, as the location illumination feature corresponding to the video image; and combining the location type feature and the location illumination feature of the video image into the environmental feature of the video image.

[0042] Specifically, the location type can be determined based on the installation location of the video surveillance camera to which the video image belongs. Location types can include entrances / exits, surface parking lots, underground parking lots, underground equipment rooms, etc. Each location type corresponds to a location type feature, which is a numerical value. Since a location type corresponds to one or more installation locations, the corresponding location type can be determined based on the installation location of the video monitor. The location type feature can then be obtained based on the location type corresponding to each video image.

[0043] Furthermore, based on the current time point and the installation location corresponding to the video image, matching location lighting features are obtained from the environmental feature acquisition rules. These location lighting features are used to characterize the lighting at the location where the video image was acquired. Specifically, the lighting type of the installation location is obtained. The lighting types configured in the environmental feature acquisition rules include enclosed, semi-enclosed, and open types. Each lighting type corresponds to one or more location types. Enclosed types include underground parking lots, underground equipment rooms, etc.; semi-enclosed types include entrances and exits, etc.; and open types include surface parking lots, etc. The ambient light intensity is determined based on the current time. Different times correspond to different sunlight intensities, resulting in different ambient light intensities. The environmental feature acquisition rules also include the ambient light intensity corresponding to different time points. For example, if the current time is 8:00 to 16:29, the ambient light intensity is determined to be 10; if the current time is 16:30 to 17:59, the ambient light intensity is determined to be 8; if the current time is 18:00 to 19:29, the ambient light intensity is determined to be 5; and if the current time is 19:30 to 5:29 the next day, the ambient light intensity is determined to be 2. If the lighting type of the installation site is enclosed, the corresponding site lighting feature is determined to be "1"; if the lighting type of the installation site is semi-enclosed, the corresponding site lighting feature is determined to be [r / 2+1], where r is the ambient light intensity corresponding to the current time, and [x] is the rounding symbol for x. If the lighting type of the installation site is open, then the lighting characteristics of the installation site are determined to be the ambient light intensity at the current time.

[0044] By combining the location type features and location lighting features of a video image, the environmental features of the video image can be obtained.

[0045] In a specific embodiment, after combining the location type feature and location lighting feature of the video image into the environmental feature of the video image, the method further includes: obtaining weather information corresponding to the current time point; obtaining weather features that match the weather information from the environmental feature acquisition rules; determining whether the installation location corresponding to the video image is an open location; if the installation location of the video image is an open location, adding the weather feature to the environmental feature of the video image; if the installation location of the video image is not an open location, adding pre-stored filler features to the environmental feature of the video image.

[0046] Furthermore, to enrich the environmental features of the video image, weather information corresponding to the current time point can be obtained. The environmental feature acquisition rules also include weather features matching each piece of weather information; for example, sunny weather corresponds to a weather feature of "10", cloudy weather to "8", overcast weather to "6", rainy weather to "4", and snowy weather to "2". Further, it is determined whether the installation location corresponding to the video image is an open location. If it is an open location, the weather features are added to the environmental features of the video image. If the installation location is not an open location, a padding feature is added to the environmental features of the video image. For example, the padding feature could be "0".

[0047] S140. Combine the image features and environmental features of each video image and input them into a preset anomaly recognition model to perform anomaly recognition on each video image and obtain the corresponding monitoring recognition result.

[0048] The image features and environmental features of each video image are combined and input into a preset anomaly recognition model to perform anomaly recognition on each video image and obtain the corresponding monitoring recognition result. Furthermore, the image features and environmental features of the video images can be combined, and the combined data information can be input into the anomaly recognition model; a set of data information corresponding to each video image is input into the anomaly recognition model, and the anomaly recognition model can perform anomaly recognition on a set of data information for each video image, thereby obtaining the monitoring recognition result.

[0049] In a specific embodiment, step S140 specifically includes the following sub-steps: combining the image features with the environmental features to obtain combined features; inputting the feature values ​​contained in the combined features to the corresponding input node in the anomaly recognition model, so as to obtain the corresponding anomaly probability value as the corresponding monitoring and recognition result through the anomaly recognition model.

[0050] Specifically, image features and environmental features can be combined. Image features are a 1×M dimensional array (e.g., a 1×256 dimensional array), and environmental features consist of multiple feature values. The combination yields a 1×(M+N) array, which is the combined feature, where N is the number of feature values ​​in the environmental features. The feature values ​​contained in the combined feature are input to the input nodes of the anomaly detection model, so the number of feature values ​​in the combined feature is equal to the number of input nodes. The anomaly detection model includes an intermediate layer and an output node. The intermediate layer contains at least one layer of intermediate nodes, which are arranged in a series of intermediate nodes. The intermediate layer performs correlation analysis on the feature values ​​input from the input nodes, and the output node outputs the corresponding anomaly probability value. The anomaly probability value is the probability of an anomaly existing in the video image; a higher anomaly probability value indicates a greater likelihood of an anomaly in the video image.

[0051] In a specific embodiment, before combining the image features and environmental features of each video image and inputting them into a preset anomaly recognition model, the method further includes: training the initial anomaly recognition model according to multiple preset training image sets to obtain candidate recognition models corresponding to each training image set; verifying each candidate recognition model according to a preset verification image set to obtain the accuracy of each candidate recognition model; and selecting the candidate recognition model with the highest accuracy as the trained anomaly recognition model.

[0052] Specifically, before using the anomaly detection model for anomaly analysis, an initial anomaly detection model can be trained using multiple pre-set training image sets. This initial anomaly detection model is essentially an untrained AI model, resulting in only one initial model. The initial anomaly detection model can be trained separately using multiple training image sets, each containing multiple training images. The number of training images in each training image set can be equal or unequal. Each training image corresponds to a set of combined features and an anomaly label. The combined features of the training images in the training image set are sequentially input into the initial anomaly detection model, and the output value of the output node is obtained. Gradient descent is then performed using the anomaly label of the training image combined with the output value of the output node. This gradient descent operation is used to fine-tune the parameter values ​​in the initial anomaly detection model; this parameter fine-tuning process is equivalent to training the initial anomaly detection model. The initial anomaly detection model can be trained once using a single training image, and iteratively trained using multiple training images from the training image set. After each training image in the training image set has been used to train the initial anomaly detection model, a candidate detection model corresponding to that training image set can be obtained. Each training image set can then yield a corresponding candidate recognition model.

[0053] Each candidate recognition model is validated using a set of validation images, which contains multiple validation images. Each validation image corresponds to a set of combined features and an anomaly label. Specifically, the combined features of the validation images in the set are sequentially input into the candidate recognition models. The models then obtain the output results for each validation image and determine whether these output results match the anomaly label. The ratio of the number of output results matching the anomaly label to the total number of validation images is used as the accuracy of each candidate recognition model. Each candidate image can then have a corresponding accuracy rate. The candidate recognition model with the highest accuracy is selected as the trained anomaly recognition model for use.

[0054] S150. Determine whether there are any abnormalities in the monitoring and recognition results of each video image.

[0055] Determine whether there are any anomalies in the monitoring and recognition results of each video image. Furthermore, the anomaly probability value in the monitoring and recognition results of the video images can be determined to distinguish whether there are any anomalies in the monitoring and recognition results.

[0056] In a specific embodiment, step S150 specifically includes the step of: determining whether the abnormal probability value in the monitoring and identification result is not less than a pre-stored abnormal probability threshold, thereby determining whether the monitoring and identification result is abnormal.

[0057] Specifically, it can be determined whether the abnormal probability value in the monitoring and identification result is not less than the pre-stored abnormal probability threshold. For example, the abnormal probability threshold can be set to 60%. If the abnormal probability value in the monitoring and identification result is not less than the abnormal probability threshold, it is determined that the monitoring and identification result is abnormal. If the abnormal probability value in the monitoring and identification result is less than the abnormal probability threshold, it is determined that the monitoring and identification result is not abnormal.

[0058] S160. If the monitoring and recognition results of each video image are normal, return to the step of extracting the corresponding video image from the video stream input by the video monitor according to the preset parking monitoring cycle.

[0059] If the monitoring and identification results of all the video images are normal, return to the step of extracting the corresponding video image from the video stream input by the video monitor according to the preset parking monitoring cycle. If the monitoring and identification results of all video images are normal, then extract video images from the video stream again according to the parking monitoring cycle for anomaly identification, that is, return to step S110.

[0060] S170. If any of the video images shows an abnormality in the monitoring and recognition result, a corresponding alarm message shall be issued.

[0061] If any video image monitoring result shows an anomaly, a corresponding alarm message will be generated based on the installation location of the video monitor corresponding to the video image. The alarm message includes the installation location of the video image and the time the message was generated. Users can view the alarm message to obtain the specific location of the anomaly (i.e., the installation location of the video image) and the time when the anomaly was discovered (i.e., the time the message was generated).

[0062] In a specific embodiment, after step S170, the method further includes the steps of: generating an exception handling work order corresponding to the monitoring identification result; and storing the exception handling work order in a pre-configured monitoring work order table.

[0063] Furthermore, after issuing an alarm notification, a corresponding anomaly handling work order can be generated based on the monitoring identification results. This work order records the occurrence, processing status, and subsequent handling operations of the anomaly. The generated work order can be dispatched to property management personnel, who can then handle the anomaly based on it. To further consolidate and record these work orders, improving the efficiency of viewing and processing them, the generated work orders can be stored in a pre-configured monitoring work order table. Users can then quickly view the processing progress and handler information for each work order by checking the monitoring work order table, thus improving the efficiency of viewing, processing, and statistically analyzing anomaly handling work orders.

[0064] The parking lot monitoring method based on image recognition provided in this embodiment of the invention extracts video images from the video stream according to the parking monitoring cycle and extracts image features to obtain environmental features corresponding to the video images. The environmental features and image features are then merged and input into an anomaly recognition model to obtain the monitoring and recognition result of the video images. The method determines whether the monitoring and recognition result is abnormal. If there is no abnormality, the step of extracting video images continues; if there is an abnormality, an alarm message is issued. This parking lot monitoring method can utilize the video stream of existing video monitors to extract video images for monitoring and analysis. Therefore, it is not limited by the display size of the video screen and can analyze clearer video images extracted from the original video stream. This significantly improves the efficiency and reliability of anomaly monitoring of vehicles and charging facilities in parking lots, and can accurately monitor abnormal problems occurring in parking lots using the video stream of video monitors.

[0065] This invention also provides an image recognition-based parking lot monitoring device, which can be configured in a management server and is used to execute any of the aforementioned image recognition-based parking lot monitoring methods. Specifically, please refer to FIG3, which is a schematic block diagram of the image recognition-based parking lot monitoring device provided in this invention embodiment.

[0066] As shown in Figure 3, the parking lot monitoring device 100 based on image recognition includes a video image capture unit 110, an image feature extraction unit 120, an environmental feature acquisition unit 130, a monitoring and recognition result acquisition unit 140, an anomaly judgment unit 150, a return execution unit 160, and an alarm prompting unit 170.

[0067] The video image capture unit 110 is used to capture corresponding video images from the video stream input by the video monitor according to a preset parking monitoring cycle.

[0068] The image feature extraction unit 120 is used to extract corresponding image features from each of the video images according to a preset image feature extraction model.

[0069] The environmental feature acquisition unit 130 is used to acquire the environmental features corresponding to the video monitor to which each video image belongs, according to the preset environmental feature acquisition rules.

[0070] The monitoring and identification result acquisition unit 140 is used to combine the image features and environmental features of each video image and input them into a preset anomaly identification model to perform anomaly identification on each video image and obtain the corresponding monitoring and identification result.

[0071] The anomaly detection unit 150 is used to determine whether there are any anomalies in the monitoring and recognition results of each video image.

[0072] The execution unit 160 is used to return to the step of extracting the corresponding video image from the video stream input by the video monitor according to the preset parking monitoring cycle if the monitoring and recognition results of each video image are normal.

[0073] The alarm notification unit 170 is used to issue a corresponding alarm notification message if the monitoring and identification result of any of the video images is abnormal.

[0074] The parking lot monitoring device based on image recognition provided in this embodiment of the invention applies the above-mentioned parking lot monitoring method based on image recognition. The method includes: extracting video images from the video stream according to the parking monitoring cycle and extracting image features; obtaining environmental features corresponding to the video images; merging the environmental features and image features and inputting them into an anomaly recognition model to obtain the monitoring and recognition result of the video images; determining whether the monitoring and recognition result is abnormal; if there is no abnormality, continuing to execute the step of extracting video images; if there is an abnormality, issuing an alarm prompt message. The above-mentioned parking lot monitoring method can utilize the video stream of existing video monitors to extract video images for monitoring and analysis, thus not being limited by the display size of the video screen, and can analyze clearer video images extracted from the original video stream, greatly improving the efficiency and reliability of anomaly monitoring of vehicles and charging facilities in parking lots, and can accurately monitor abnormal problems occurring in parking lots using the video stream of video monitors.

[0075] The aforementioned image recognition-based parking monitoring device can be implemented as a computer program, which can run on the computer device shown in Figure 4.

[0076] Please refer to Figure 4, which is a schematic block diagram of a computer device provided in an embodiment of the present invention. This computer device may be a management server for executing an image recognition-based parking lot monitoring method to acquire video streams and monitor for anomalies.

[0077] Referring to Figure 4, the computer device 500 includes a processor 502, a memory, and a communication interface 505 connected via a communication bus 501. The memory may include a storage medium 503 and internal memory 504.

[0078] The storage medium 503 may store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, it causes the processor 502 to execute a parking lot monitoring method based on image recognition. The storage medium 503 may be a volatile storage medium or a non-volatile storage medium.

[0079] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0080] The internal memory 504 provides an environment for the operation of the computer program 5032 in the storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a parking lot monitoring method based on image recognition.

[0081] The communication interface 505 is used for network communication, such as providing data information transmission. Those skilled in the art will understand that the structure shown in Figure 4 is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the computer device 500 to which the present invention is applied. A specific computer device 500 may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0082] The processor 502 is used to run the computer program 5032 stored in the memory to implement the corresponding functions in the above-described image recognition-based parking lot monitoring method.

[0083] Those skilled in the art will understand that the embodiments of the computer device shown in FIG4 do not constitute a limitation on the specific configuration of the computer device. In other embodiments, the computer device may include more or fewer components than shown, or combine certain components, or have different component arrangements. For example, in some embodiments, the computer device may include only a memory and a processor. In such embodiments, the structure and function of the memory and processor are consistent with those shown in FIG4, and will not be repeated here.

[0084] It should be understood that, in this embodiment of the invention, the processor 502 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0085] In another embodiment of the invention, a computer-readable storage medium is provided. This computer-readable storage medium may be volatile or non-volatile. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps included in the image recognition-based parking lot monitoring method described above.

[0086] Those skilled in the art will readily understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.

[0087] In the embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Units with the same function may be grouped into one unit. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, or it may be an electrical, mechanical, or other form of connection.

[0088] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention, depending on actual needs.

[0089] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0090] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a computer-readable storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned computer-readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), magnetic disks, or optical disks.

[0091] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A parking lot monitoring method based on image recognition, wherein the method is applied in a management server, the management server establishing a network connection with at least one video surveillance camera to achieve data information transmission, characterized in that, The method includes: extracting corresponding video images from the video stream input by the video monitor according to a preset parking monitoring cycle; extracting corresponding image features from each video image according to a preset image feature extraction model; obtaining environmental features corresponding to the video monitor to which each video image belongs according to preset environmental feature acquisition rules; combining the image features and environmental features of each video image and inputting them into a preset anomaly recognition model to perform anomaly recognition on each video image to obtain corresponding monitoring recognition results; the anomaly recognition model is an artificial intelligence model composed of an input node, an intermediate layer, and an output node, wherein the input node is used for... The input consists of combined features formed by image features and environmental features. The intermediate layer performs correlation analysis on the feature values ​​input from the input nodes, and the output node outputs the corresponding anomaly probability value, which reflects the probability of anomalies in the video images. The system determines whether the monitoring and recognition results of each video image are abnormal. If the monitoring and recognition results of all video images are normal, the system returns to the step of extracting the corresponding video image from the video stream input from the video monitor according to a preset parking monitoring cycle. If the monitoring and recognition result of any video image is abnormal, a corresponding alarm message is issued. The step of obtaining the preset environmental features... The rules for obtaining environmental features corresponding to the video surveillance cameras to which each video image belongs include: determining the corresponding location type feature based on the installation location of the video surveillance camera to which the video image belongs; the location type feature is a numerical value corresponding to the location type of the installation location; obtaining matching lighting features from the environmental feature acquisition rules based on the current time point and the installation location corresponding to the video image, as the location lighting feature corresponding to the video image, including: obtaining the lighting type matching the installation location from the environmental feature acquisition rules, and determining the external ambient light intensity based on the current time point; if the lighting type of the installation location is enclosed... The location type is determined as "1" for the location lighting feature corresponding to the installation location; if the location lighting type is semi-enclosed, the location lighting feature corresponding to the installation location is determined as [r / 2+1], where r is the ambient light intensity at the current time point, and [] is the rounding up symbol; if the location lighting type is open, the location lighting feature corresponding to the installation location is determined as the ambient light intensity at the current time point; the lighting type configured in the environmental feature acquisition rules includes enclosed, semi-enclosed, and open types; the location type feature and location lighting feature of the video image are combined to form the environmental feature of the video image.

2. The parking lot monitoring method based on image recognition according to claim 1, characterized in that, After combining the location type features and location lighting features of the video image into the environmental features of the video image, the method further includes: obtaining weather information corresponding to the current time point; obtaining weather features that match the weather information from the environmental feature acquisition rules; determining whether the installation location corresponding to the video image is an open location; if the installation location of the video image is an open location, adding the weather features to the environmental features of the video image; if the installation location of the video image is not an open location, adding pre-stored filler features to the environmental features of the video image.

3. The parking lot monitoring method based on image recognition according to claim 1 or 2, characterized in that, The step of combining the image features and environmental features of each video image and inputting them into a preset anomaly recognition model to perform anomaly recognition on each video image and obtain the corresponding monitoring recognition result includes: combining the image features and the environmental features to obtain combined features; inputting the feature values ​​contained in the combined features into the corresponding input node in the anomaly recognition model to perform anomaly recognition through the anomaly recognition model and obtain the corresponding anomaly probability value as the corresponding monitoring recognition result.

4. The parking lot monitoring method based on image recognition according to claim 3, characterized in that, Before combining the image features and environmental features of each video image and inputting them into a preset anomaly recognition model, the method further includes: training the initial anomaly recognition model according to multiple preset training image sets to obtain candidate recognition models corresponding to each training image set; verifying each candidate recognition model according to a preset verification image set to obtain the accuracy of each candidate recognition model; and selecting the candidate recognition model with the highest accuracy as the trained anomaly recognition model.

5. The parking lot monitoring method based on image recognition according to claim 1 or 2, characterized in that, The step of determining whether the monitoring and identification results of each video image are abnormal includes: determining whether the abnormal probability value in the monitoring and identification results is not less than a pre-stored abnormal probability threshold, thereby determining whether the monitoring and identification results are abnormal.

6. The parking lot monitoring method based on image recognition according to claim 5, characterized in that, After issuing the corresponding alarm message, the method further includes: generating an anomaly handling work order corresponding to the monitoring identification result; and storing the anomaly handling work order in a pre-configured monitoring work order table.

7. A parking lot monitoring device based on image recognition, characterized in that, The device is configured in a management server, which establishes a network connection with at least one video monitor to achieve data transmission. The device applies the parking lot monitoring method based on image recognition as described in any one of claims 1-6. The device includes: a video image capture unit, used to capture corresponding video images from the video stream input by the video monitor according to a preset parking monitoring cycle; an image feature extraction unit, used to extract corresponding image features from each of the video images according to a preset image feature extraction model; and an environmental feature acquisition unit, used to acquire the corresponding video monitor to which each of the video images belongs according to preset environmental feature acquisition rules. The system includes: an environmental feature detection unit; a monitoring and identification result acquisition unit, used to combine the image features and environmental features of each video image and input them into a preset anomaly identification model to perform anomaly identification on each video image and obtain the corresponding monitoring and identification result; an anomaly judgment unit, used to determine whether there is an anomaly in the monitoring and identification result of each video image; a return execution unit, used to return to the step of extracting the corresponding video image from the video stream input by the video monitor according to the preset parking monitoring cycle if there is no anomaly in the monitoring and identification result of each video image; and an alarm prompting unit, used to issue a corresponding alarm prompting message if there is an anomaly in the monitoring and identification result of any video image.

8. A computer device, characterized in that, The device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory is used to store computer programs; and the processor, when executing the program stored in the memory, implements the steps of the parking lot monitoring method based on image recognition as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the parking lot monitoring method based on image recognition as described in any one of claims 1-6.

Images