A vehicle-mounted camera anomaly detection method, device, equipment and storage medium

Abstract

The application discloses a vehicle-mounted camera anomaly detection method and device, equipment and a storage medium, relates to the field of vehicle engineering, and is applied to a target cloud and comprises the following steps: sending an anomaly detection instruction to a target vehicle terminal of a target vehicle based on a preset trigger condition to obtain a target media file; analyzing a target statistical characteristic of the target media file to obtain a first analysis result, determining a target detection model based on a target camera corresponding to the target media file, and performing reference area checking on the target media file by using the target detection model to obtain a corresponding second analysis result; determining a first confidence degree corresponding to the first analysis result and a second confidence degree corresponding to the second analysis result by using a target confidence degree algorithm, and fusing the first analysis result and the second analysis result based on the first confidence degree and the second confidence degree to obtain an anomaly detection result corresponding to the target camera. The application improves the accuracy and efficiency of vehicle-mounted camera anomaly detection.

Description

Technical Field

[0001] This invention relates to the field of vehicle engineering, and in particular to a method, apparatus, equipment, and storage medium for detecting anomalies in vehicle-mounted cameras. Background Technology

[0002] With the increasing intelligence and informatization of vehicles, various camera devices are commonly installed for driving assistance, safety monitoring, and operation management. Commercial vehicle safety monitoring and risk management service personnel need to continuously and frequently monitor the data from these cameras to complete safety risk control services. However, due to improper installation by personnel, driver-intentioned or unintentional obstruction, and long-term operation, cameras are susceptible to damage from dust, dirt, aging, and other factors, leading to decreased video quality or even complete unusability. Ultimately, this affects the ability of safety monitoring and risk management service personnel to provide continuous safety assurance services to drivers. Summary of the Invention

[0003] In view of this, the purpose of this invention is to provide a method, apparatus, device, and storage medium for detecting anomalies in vehicle-mounted cameras, which can improve the accuracy and efficiency of anomaly detection in vehicle-mounted cameras. The specific solution is as follows: In a first aspect, this application discloses an anomaly detection method for vehicle-mounted cameras, applied to a target cloud, comprising: An anomaly detection command is sent to the target vehicle's infotainment system based on preset trigger conditions to obtain the target media file collected by the target vehicle's infotainment system using the target camera based on the anomaly detection command; the target media file includes an image to be detected and a video segment to be detected. The target statistical characteristics of the target media file are analyzed to obtain the corresponding first analysis result. Based on the target camera corresponding to the target media file, a target detection model is determined. Using the target detection model, the target media file is verified based on the target reference area corresponding to the target camera to obtain the corresponding second analysis result. The first confidence level corresponding to the first analysis result and the second confidence level corresponding to the second analysis result are determined using a target confidence algorithm. The first analysis result and the second analysis result are then fused based on the first confidence level and the second confidence level to obtain the anomaly detection result corresponding to the target camera.

[0004] Optionally, before sending the anomaly detection command to the target vehicle's infotainment system based on preset triggering conditions, the method further includes: Determine the preset trigger conditions; The preset triggering conditions include a first triggering condition, a second triggering condition, and a third triggering condition; the first triggering condition is a preset triggering condition set based on a preset detection cycle; the second triggering condition is a preset triggering condition determined based on the real-time operating status of the target vehicle; and the third triggering condition is a preset triggering condition determined based on the historical anomaly detection results of the target vehicle.

[0005] Optionally, the step of analyzing the target statistical characteristics of the target media file to obtain the corresponding first analysis result includes: Based on the device type, installation environment, and historical operating data of the target camera, a target threshold is determined for the target camera; the target threshold includes a grayscale average threshold, a grayscale variance threshold, and a brightness variation threshold. Based on the target threshold, the target statistical characteristics of the target media file corresponding to the target camera are analyzed to obtain the corresponding first analysis result.

[0006] Optionally, before performing reference region verification on the target media file based on the target reference region corresponding to the target camera, the method further includes: Determine the target reference area corresponding to the target camera based on the device type of the target camera; Wherein, if the target camera is an interior camera of the vehicle, the target reference area includes the steering wheel and the human face; if the target camera is an exterior camera of the vehicle, the target reference area includes the driving road and lane lines.

[0007] Optionally, the step of using the target detection model to perform reference region verification on the target media file based on the target reference region corresponding to the target camera to obtain the corresponding second analysis result includes: If the target camera is an interior camera used to capture images of the occupants of the target vehicle, then the first detection model is used to perform a first reference region verification on the target media file based on the target reference region; If the first reference region verification of the target media file is completed within the preset detection period, the corresponding second analysis result is obtained based on the first reference region verification result of the target media file. If the first reference region verification of the target media file is not completed within the preset detection period, the second detection model is used to perform a second reference region verification on the target media file based on the target reference region to obtain the corresponding second reference region verification result, and the corresponding second analysis result is obtained based on the second reference region verification result. The first detection model is a detection model that determines the human facial structure by recognizing key facial feature points, and the second detection model is a YOLO model used to recognize the human head; the detection efficiency of the first detection model is higher than that of the second detection model, and the detection accuracy of the second detection model is higher than that of the first detection model.

[0008] Optionally, after obtaining the anomaly detection result corresponding to the target camera, the method further includes: The anomaly detection results are saved to the target storage space, and all the anomaly detection results of the target vehicle within the preset time window are obtained from the target storage space based on the preset trend analysis period, so as to determine the target anomaly trend of the target camera of the target vehicle within the preset time window based on the anomaly detection results using preset trend analysis rules.

[0009] Optionally, the method for detecting anomalies in the vehicle-mounted camera further includes: The preset triggering conditions are adjusted based on the target anomaly trend to obtain the adjusted preset triggering conditions, and then the process jumps to the step of sending an anomaly detection command to the target vehicle's infotainment system based on the preset triggering conditions.

[0010] Secondly, this application discloses an anomaly detection device for an in-vehicle camera, applied to a target cloud, comprising: The media file acquisition module is used to send an anomaly detection command to the target vehicle's infotainment system based on preset trigger conditions, so as to acquire the target media file collected by the target vehicle's infotainment system using the target camera based on the anomaly detection command; the target media file includes an image to be detected and a video segment to be detected. The media file analysis module is used to analyze the target statistical characteristics of the target media file to obtain the corresponding first analysis result, and to determine the target detection model based on the target camera corresponding to the target media file. Using the target detection model, the target media file is used to perform reference region verification based on the target reference region corresponding to the target camera to obtain the corresponding second analysis result. The detection result acquisition module is used to determine the first confidence level corresponding to the first analysis result and the second confidence level corresponding to the second analysis result using a target confidence algorithm, and to fuse the first analysis result and the second analysis result based on the first confidence level and the second confidence level to obtain the anomaly detection result corresponding to the target camera.

[0011] Thirdly, this application discloses an electronic device, including: Memory, used to store computer programs; A processor is used to execute the computer program to implement the aforementioned method for detecting anomalies in vehicle-mounted cameras.

[0012] Fourthly, this application discloses a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the aforementioned method for detecting anomalies in vehicle-mounted cameras.

[0013] In this application, when performing anomaly detection on an in-vehicle camera, the target cloud sends an anomaly detection command to the target vehicle's infotainment system based on preset trigger conditions to obtain target media files collected by the target vehicle's infotainment system using the target camera based on the anomaly detection command. The target media files include images to be detected and video segments to be detected. The target statistical characteristics of the target media files are analyzed to obtain corresponding first analysis results, and a target detection model is determined based on the target camera corresponding to the target media files. Using the target detection model, a reference region verification is performed on the target media files based on the target reference region corresponding to the target camera to obtain corresponding second analysis results. A target confidence algorithm is used to determine the first confidence level corresponding to the first analysis results and the second confidence level corresponding to the second analysis results. The first analysis results and the second analysis results are fused based on the first confidence level and the second confidence level to obtain the anomaly detection results corresponding to the target camera. As can be seen, this application utilizes a target cloud platform to control the target vehicle's infotainment system to collect and upload target media files based on preset trigger conditions. The target cloud platform then analyzes these media files, centralizing the main processing logic in the cloud. This reduces the processing complexity of the target vehicle's infotainment system, improves anomaly detection efficiency, and is suitable for unified management and maintenance of large-scale equipment. When performing anomaly detection based on target media files, the target cloud platform not only analyzes the target statistical characteristics of the media files but also uses the target detection model corresponding to the target camera to perform reference region verification on the corresponding target reference area. Based on confidence levels, it fuses the first and second analysis results to obtain the anomaly detection result corresponding to the target camera. Based on a combination of multi-dimensional criteria, it determines the abnormal or malfunctioning state of the device, improving the accuracy of anomaly detection for vehicle-mounted cameras. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0015] Figure 1 This is a flowchart of an anomaly detection method for a vehicle-mounted camera disclosed in this application; Figure 2 This is a schematic diagram of the structure of an anomaly detection device for a vehicle-mounted camera disclosed in this application; Figure 3 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0016] 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 embodiments of the present invention, and not all embodiments. 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.

[0017] With the increasing intelligence and informatization of vehicles, various camera devices are commonly installed on vehicles for driving assistance, safety monitoring, and operation management. Commercial vehicle safety monitoring and risk control service personnel need to continuously and frequently monitor the data fed back by these cameras to complete safety risk control services. However, due to improper installation by personnel, driver-intentioned or unintentional obstruction, and long-term operation, cameras are easily affected by dust, dirt, aging, and other factors, leading to a decline in output video quality or even complete unusability. This ultimately affects the ability of safety monitoring and risk control service personnel to provide continuous safety assurance services to drivers. To solve the above technical problems, this application discloses a method for detecting anomalies in vehicle-mounted cameras, which can improve the accuracy and efficiency of anomaly detection.

[0018] See Figure 1 As shown, this invention discloses an anomaly detection method for vehicle-mounted cameras, applied to a target cloud, comprising: Step S11: Send an anomaly detection command to the target vehicle's infotainment system based on preset trigger conditions to obtain the target media file collected by the target vehicle's infotainment system using the target camera based on the anomaly detection command; the target media file includes the image to be detected and the video segment to be detected.

[0019] In this embodiment, before sending an anomaly detection command to the target vehicle's in-vehicle system based on preset trigger conditions, the method further includes: determining preset trigger conditions; the preset trigger conditions include a first trigger condition, a second trigger condition, and a third trigger condition; the first trigger condition is a preset trigger condition set based on a preset detection cycle; the second trigger condition is a preset trigger condition determined based on the real-time operating status of the target vehicle; and the third trigger condition is a preset trigger condition determined based on the historical anomaly detection results of the target vehicle. In a specific implementation, the target cloud presets multiple trigger conditions for vehicle video or image reporting based on the characteristics of the target vehicle and the scene. These preset trigger conditions include, but are not limited to: 1. Periodic health check trigger conditions (i.e., the first trigger condition): the camera status is checked at fixed intervals, such as daily or hourly.

[0020] 2. Triggering conditions during equipment initialization or stable operation (i.e., the second triggering condition): the check is triggered when the vehicle goes online, or when the vehicle speed is greater than the preset speed.

[0021] 3. Based on the specific time window triggering conditions configured by the system policy (i.e., the third triggering condition), triggering conditions are formed by specifying multiple combination rules and combining various information such as historical inspection records to meet different inspection frequencies and quality requirements. Then, multiple cameras on the vehicle are inspected in batches.

[0022] When preset trigger conditions are met, the target cloud sends corresponding anomaly detection commands to the target vehicle's infotainment system. The infotainment system, based on these commands, uses the target camera to capture images or videos as target media files and uploads the detected video or image data to the cloud. For example, in a commercial vehicle camera status inspection example, the cloud platform sets the following combined conditions: The cloud platform continuously monitors the trajectory data reported by the vehicle's devices. When the vehicle remains online and at a speed greater than 10 km / h for more than 5 minutes, the driver is considered to be in a stable driving state. At this time, the platform simultaneously queries the historical inspection results recorded by the vehicle's camera platform. If the vehicle's camera inspection results are normal within 24 hours, no repeated inspection is performed to reduce bandwidth consumption; if the previous inspection status was abnormal, a re-inspection is triggered every hour, sending an upload command to the terminal device. It is understood that the above preset trigger conditions can be fixed or dynamically adjusted based on previous inspection results.

[0023] Step S12: Analyze the target statistical characteristics of the target media file to obtain the corresponding first analysis result, and determine the target detection model based on the target camera corresponding to the target media file. Using the target detection model, perform reference region verification on the target media file based on the target reference region corresponding to the target camera to obtain the corresponding second analysis result.

[0024] In this embodiment, after receiving the reported video or image, the target cloud performs storage and availability assessment on the target media file. First, the video file is read to obtain its size, dimensions, and encoding information. It is then opened frame by frame, converted into image files, and decoded to determine if the target media file has been completely decoded. For example, when using the OpenCV framework to read the video, the video size, total number of frames, resolution, and frame rate (FPS) are first obtained. If the video cannot be opened or parsing fails, it is considered corrupted. Then, each image frame is read and processed to confirm that there are no corrupted frames in the video.

[0025] In this embodiment, if the target media file can be fully decoded, the target statistical characteristics of the target media file are analyzed to obtain a corresponding first analysis result. This includes: determining a target threshold for the target camera based on the device type, installation environment, and historical operating data corresponding to the target camera; the target threshold includes a grayscale average threshold, a grayscale variance threshold, and a brightness variation threshold; and analyzing the target statistical characteristics of the target media file corresponding to the target camera based on the target threshold to obtain the corresponding first analysis result. In a specific implementation, the camera can be judged to be in an unreasonable operating state by analyzing various basic statistical characteristics of the video, including but not limited to brightness and contrast. Based on the device type, installation environment, and historical operating data, a reasonable range is evaluated, and a pre-set threshold range for each statistical characteristic is used to obtain the first analysis result.

[0026] In one specific implementation, the average grayscale value of each frame in the video is calculated, and the average grayscale value of the entire video segment is summed to obtain the average video brightness. A preset threshold range is set. When the average video brightness is below the minimum value, the video is considered abnormally dark, possibly indicating a black screen or severe obstruction. When the average video brightness is above the maximum value, the video is considered abnormally bright, possibly indicating overexposure or strong reflection obstruction. At this point, the concentration of brightness peaks is calculated using a grayscale histogram. If the variance of the peak values ​​in each frame of the video is too small, it indicates that the video brightness is concentrated in a narrow range, carrying insufficient information; thus distinguishing between a completely black screen, an overexposed white screen, and normal video images. The contrast level of the video is evaluated by statistically analyzing the variance of pixel grayscale values ​​in the video frames. If the variance is too low, the video is likely too blurry or covered by dirt. If the variance is too high, there is extreme noise or abnormal lighting. Furthermore, the trend of brightness variation in consecutive video frames is calculated. If the brightness variation is consistently below a preset threshold, it indicates that the video is in an abnormally stable brightness state, thus distinguishing and identifying camera failure caused by stickers or fixed obstructions. Generally speaking, if the brightness is normal but the contrast is consistently low, it can be judged as a possible obstruction by dirt or stains. If both the brightness and contrast are extremely low, it can be judged as a black screen or complete obstruction. The first analysis result obtained by combining multiple indicators through engineering methods can effectively identify abnormal failures of a single video or image from the camera, such as black screen, obstruction, stickers, dirt, white screen, overexposure, and reflection.

[0027] In this embodiment, before performing reference region verification on the target media file based on the target reference region corresponding to the target camera, the method further includes: determining the target reference region corresponding to the target camera based on the device type of the target camera; wherein, if the target camera is an interior camera of a vehicle, the target reference region includes the steering wheel and human face; if the target camera is an exterior camera of a vehicle, the target reference region includes the driving road and lane lines. The target reference region refers to an image region or structural feature region that exists stably and continuously in the video frame after the device is installed and in normal operating condition. The types of preset target reference regions vary and are selectable in different scenarios. In the video availability assessment process, this embodiment selects a structural region that exists stably and continuously in the video frame under normal installation condition as the target reference region, and performs a consistency judgment on the existence and structural characteristics of the target reference region to obtain the corresponding second analysis result. When the target reference region is continuously missing or deviates from the preset structural characteristic range in multiple video reporting cycles, the device output capability is determined to be abnormal. For example, in commercial vehicles, the DMS (Driver Monitoring System) camera inside the vehicle can select the steering wheel, human face, etc. as the stable structural reference area, while for the ADAS (Advanced Driver Assistance System) camera outside the vehicle, the road, car, lane lines, etc. can be selected as the target reference area.

[0028] In one specific implementation, a target detection model is used to perform reference region verification on a target media file based on a target reference region corresponding to a target camera, in order to obtain a corresponding second analysis result. This includes: if the target camera is an interior camera used to capture images of the occupants of a target vehicle, then a first detection model is used to perform a first reference region verification on the target media file based on the target reference region; if the first reference region verification of the target media file is completed within a preset detection period, then a corresponding second analysis result is obtained based on the first reference region verification result of the target media file; if the first reference region verification of the target media file is not completed within the preset detection period, then a second detection model is used to perform a second reference region verification on the target media file based on the target reference region to obtain a corresponding second reference region verification result, and a corresponding second analysis result is obtained based on the second reference region verification result; wherein, the first detection model is a detection model that determines the human facial structure by recognizing facial feature key points, and the second detection model is a YOLO model used to recognize the human head; the detection efficiency of the first detection model is higher than that of the second detection model, and the detection accuracy of the second detection model is higher than that of the first detection model.

[0029] Taking the failure detection scenario of in-vehicle DMS cameras in commercial vehicles as an example, using the human head and face as the target reference area, after comparing the engineering performance of relevant algorithms, the first step is to quickly identify key facial features to determine typical and clear human facial structural features. For example, the FaceMash model (i.e., the first detection model) is used to determine the face structure LandMark feature set. This feature set is used to determine whether a human face exists and the rectangular range of the face's location, including information such as the positions of the eyes, mouth, and nose tip. The platform pre-sets the range value of the calibration box, such as 80%, as the region's out-of-limit threshold; areas where the center of the image exceeds the threshold in all directions are considered normal and valid areas. First, it is determined whether the rectangular range of the small fixed structure reference area of ​​the "eyes, nose, and mouth" of the relatively accurate face exceeds the out-of-limit threshold range, and a clear out-of-limit offset failure judgment is performed to obtain the first reference area verification result. If the facial feature model (i.e., the first detection model) fails to quickly identify the human face and corresponding location, a more accurate but slightly slower human head recognition model (i.e., the second detection model) is further employed to determine the structural features of the human head and obtain a second reference region verification result. For example, a YOLO model trained with specific enhancements can be used to determine the rectangular range of the head or face for further judgment to obtain the corresponding second analysis result. If the range of the fixed head structure reference region exceeds the out-of-limit threshold, it is also considered a low-confidence single out-of-limit offset failure judgment result and participates in subsequent long-term failure checks.

[0030] Step S13: Use the target confidence algorithm to determine the first confidence level corresponding to the first analysis result and the second confidence level corresponding to the second analysis result, and fuse the first analysis result and the second analysis result based on the first confidence level and the second confidence level to obtain the anomaly detection result corresponding to the target camera.

[0031] In this embodiment, after obtaining the first analysis result and the second analysis result, a first confidence level corresponding to the first analysis result is determined based on the target confidence algorithm, and a second confidence level corresponding to the second analysis result is also determined. Then, the first analysis result and the second analysis result are fused based on the first confidence level and the second confidence level to obtain the anomaly detection result corresponding to the target camera. In a specific implementation, the multi-dimensional analysis results mentioned in the preceding steps can be summarized and judged according to a certain order or weight and confidence level to maximize the probability of forming a single anomaly detection result. For example, anomaly detection results such as normal, black screen, occlusion, high confidence offset failure, low confidence offset failure, etc., can be output.

[0032] In this embodiment, after obtaining the anomaly detection results corresponding to the target camera, the method further includes: saving the anomaly detection results to the target storage space, and obtaining all anomaly detection results of the target vehicle within a preset time window from the target storage space based on a preset trend analysis period, so as to determine the target anomaly trend of the target camera of the target vehicle within the preset time window based on the anomaly detection results using preset trend analysis rules. That is, the target cloud can record and save relevant information for each camera failure check, including device information, check start time, check end time, camera channel number, normal or failed status, failure cause, failure degree, confidence level, etc., and perform trend analysis on the anomaly criterion data formed by the same device in multiple reporting periods based on this information to obtain the corresponding target anomaly trend, such as: the continued existence of the abnormal state; the gradual aggravation of the abnormality degree; and the abnormal state failing to recover across multiple operating cycles. Based on this cross-cycle trend analysis mechanism, misjudgment caused by instantaneous anomalies can be effectively avoided.

[0033] In one specific implementation, the results of multiple inspections at different frequencies within a specified time period are calculated using the device or vehicle number as a dimension, and can be combined with related reference data from other sources for analysis and calculation. For example, tracing back to the last time the camera was normal, calculating the interval from the current time, finding the frequency or interval of consecutive normal operation and consecutive failure of the camera within a specified time period, and the frequency, confidence level changes, and correlation of one or more failure causes within a specified time period. Furthermore, manual inspection can be combined for verification and corrective adjustments. When the target anomaly trend across cycles meets preset judgment conditions, the cloud classifies the device as a short-term or long-term abnormal or failed state and generates corresponding failure results to describe specific and clear anomaly identification information for device alarms, maintenance correction, or further accurate failure management. Specific failure results can include normal operation, device malfunction, intermittent device connection failure, long-term black screen, long-term screen distortion, long-term obstruction, intermittent obstruction, stains, abnormal installation angle, intermittent abnormal angle deflection, etc., or frequency trend correlations such as increased obstruction rate, increased camera angle offset, etc. For example, if the screen is constantly distorted, the cause could be: hardware failure / shielding grounding problem / wiring workmanship problem; if it is always fixedly obstructed, the cause could be: incorrect installation location / assembly defect; if the screen is always blurry, the cause could be: poor installation workmanship / incorrect focus / factory defect; if the obstruction is intermittent, the cause could be: swinging object / temporary obstruction during loading / intrusion of moving parts; if the screen history is normal but recent sporadic interruptions have increased, the cause could be: loose connection / poor wiring harness contact; if the screen suddenly and continuously goes black recently, the cause could be: power failure / broken wiring harness / damaged camera; if the screen interrupts during bumps, the cause could be: loose plug / broken strand; if the screen history is... If the image is normally clear but suddenly becomes permanently obstructed, the cause could be: human-caused obstruction / wiring harness displacement / accessory intrusion into the field of view. If the obstructed area gradually expands, the cause could be: mud spots / water droplet accumulation / objects sliding in / crack expansion. If the image has been clear for a long time but the blurriness has been continuously increasing recently, the cause could be: driving damage / dust / mud film accumulation. If the image becomes severely blurry and persists after a certain incident, the cause could be: foreign object obstruction / lens damage / protective cover malfunction / focus failure. If the image is only blurry in wet and cold weather, the cause could be: fogging / poor sealing. If the image is only blurry at high speeds, the cause could be: vibration blur / shutter strategy problem / resonance.

[0034] In this embodiment, the preset triggering conditions can also be adjusted based on the target anomaly trend to obtain the adjusted preset triggering conditions, and then jump to the step of sending an anomaly detection command to the target vehicle's infotainment system based on the preset triggering conditions, thereby optimizing the anomaly judgment logic and improving the accuracy of anomaly detection.

[0035] As can be seen, this application utilizes a target cloud platform to control the target vehicle's infotainment system to collect and upload target media files based on preset trigger conditions. The target cloud platform then analyzes these media files, centralizing the main processing logic in the cloud. This reduces the processing complexity of the target vehicle's infotainment system, improves anomaly detection efficiency, and is suitable for unified management and maintenance of large-scale equipment. When performing anomaly detection based on target media files, the target cloud platform not only analyzes the target statistical characteristics of the media files but also uses the target detection model corresponding to the target camera to perform reference region verification on the corresponding target reference area. Based on confidence levels, it fuses the first and second analysis results to obtain the anomaly detection result corresponding to the target camera. Based on a combination of multi-dimensional criteria, it determines the abnormal or malfunctioning state of the device, improving the accuracy of anomaly detection for vehicle-mounted cameras.

[0036] See Figure 2 As shown, this application discloses an anomaly detection device for vehicle-mounted cameras, applied to a target cloud, comprising: The media file acquisition module 11 is used to send an anomaly detection command to the target vehicle's infotainment system based on a preset trigger condition, so as to acquire the target media file collected by the target vehicle's infotainment system using the target camera based on the anomaly detection command; the target media file includes an image to be detected and a video segment to be detected. The media file analysis module 12 is used to analyze the target statistical characteristics of the target media file to obtain the corresponding first analysis result, and determine the target detection model based on the target camera corresponding to the target media file. Using the target detection model, the target media file is used to perform reference area verification based on the target reference area corresponding to the target camera to obtain the corresponding second analysis result. The detection result acquisition module 13 is used to determine the first confidence level corresponding to the first analysis result and the second confidence level corresponding to the second analysis result using the target confidence algorithm, and to fuse the first analysis result and the second analysis result based on the first confidence level and the second confidence level to obtain the anomaly detection result corresponding to the target camera.

[0037] As can be seen, this application utilizes a target cloud platform to control the target vehicle's infotainment system to collect and upload target media files based on preset trigger conditions. The target cloud platform then analyzes these media files, centralizing the main processing logic in the cloud. This reduces the processing complexity of the target vehicle's infotainment system, improves anomaly detection efficiency, and is suitable for unified management and maintenance of large-scale equipment. When performing anomaly detection based on target media files, the target cloud platform not only analyzes the target statistical characteristics of the media files but also uses the target detection model corresponding to the target camera to perform reference region verification on the corresponding target reference area. Based on confidence levels, it fuses the first and second analysis results to obtain the anomaly detection result corresponding to the target camera. Based on a combination of multi-dimensional criteria, it determines the abnormal or malfunctioning state of the device, improving the accuracy of anomaly detection for vehicle-mounted cameras.

[0038] In one specific embodiment, the device may further include: The condition determination module is used to determine the preset trigger conditions; The preset triggering conditions include a first triggering condition, a second triggering condition, and a third triggering condition; the first triggering condition is a preset triggering condition set based on a preset detection cycle; the second triggering condition is a preset triggering condition determined based on the real-time operating status of the target vehicle; and the third triggering condition is a preset triggering condition determined based on the historical anomaly detection results of the target vehicle.

[0039] In one specific implementation, the media file analysis module 12 may include: The threshold determination unit is used to determine the target threshold corresponding to the target camera based on the device type, installation environment and historical operating data of the target camera; the target threshold includes the grayscale average threshold, the grayscale variance threshold and the brightness change amplitude threshold; The first analysis unit is used to analyze the target statistical characteristics of the target media file corresponding to the target camera based on the target threshold, so as to obtain the corresponding first analysis result.

[0040] In one specific embodiment, the device may further include: A target region determination unit is used to determine a target reference region corresponding to the target camera based on the device type of the target camera; Wherein, if the target camera is an interior camera of the vehicle, the target reference area includes the steering wheel and the human face; if the target camera is an exterior camera of the vehicle, the target reference area includes the driving road and lane lines.

[0041] In one specific implementation, the media file analysis module 12 may include: The first verification unit is used to perform a first reference area verification on the target media file based on the target reference area using a first detection model if the target camera is an in-vehicle camera used to capture the driver and passengers of the target vehicle. The second analysis unit is used to obtain a corresponding second analysis result based on the first reference area verification result of the target media file if the first reference area verification of the target media file is completed within a preset detection period. The second verification unit is used to perform a second reference region verification on the target media file based on the target reference region using a second detection model if the first reference region verification of the target media file is not completed within a preset detection period, and to obtain a corresponding second reference region verification result based on the second reference region verification result. The first detection model is a detection model that determines the human facial structure by recognizing key facial feature points, and the second detection model is a YOLO model used to recognize the human head; the detection efficiency of the first detection model is higher than that of the second detection model, and the detection accuracy of the second detection model is higher than that of the first detection model.

[0042] In one specific embodiment, the device may further include: An anomaly trend analysis module is used to save the anomaly detection results to a target storage space, and to obtain all the anomaly detection results of the target vehicle within a preset time window from the target storage space based on a preset trend analysis period, so as to determine the target anomaly trend of the target camera of the target vehicle within the preset time window based on the anomaly detection results using preset trend analysis rules.

[0043] In one specific embodiment, the device may further include: The condition adjustment module is used to adjust the preset triggering conditions based on the target abnormal trend to obtain the adjusted preset triggering conditions, and then jump to the step of sending an abnormality detection command to the target vehicle's infotainment system based on the preset triggering conditions.

[0044] Furthermore, embodiments of this application also disclose an electronic device, Figure 3 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0045] Figure 3This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the vehicle-mounted camera anomaly detection method disclosed in any of the foregoing embodiments. Alternatively, the electronic device 20 in this embodiment may specifically be a computer.

[0046] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0047] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0048] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the vehicle camera anomaly detection method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include a computer program capable of performing other specific tasks.

[0049] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned disclosed method for detecting anomalies in vehicle-mounted cameras. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0050] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0051] Those skilled in the art will further 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 components and steps of the various examples have been generally described in terms of functionality 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 application.

[0052] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0053] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0054] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for detecting anomalies in a vehicle-mounted camera, characterized in that, Applied to the target cloud, including: An anomaly detection command is sent to the target vehicle's infotainment system based on preset trigger conditions to obtain the target media file collected by the target vehicle's infotainment system using the target camera based on the anomaly detection command; the target media file includes an image to be detected and a video segment to be detected. The target statistical characteristics of the target media file are analyzed to obtain the corresponding first analysis result. Based on the target camera corresponding to the target media file, a target detection model is determined. Using the target detection model, the target media file is verified based on the target reference area corresponding to the target camera to obtain the corresponding second analysis result. The first confidence level corresponding to the first analysis result and the second confidence level corresponding to the second analysis result are determined using a target confidence algorithm. The first analysis result and the second analysis result are then fused based on the first confidence level and the second confidence level to obtain the anomaly detection result corresponding to the target camera.

2. The method for detecting anomalies in a vehicle-mounted camera according to claim 1, characterized in that, Before sending the anomaly detection command to the target vehicle's infotainment system based on preset triggering conditions, the process also includes: Determine the preset trigger conditions; The preset triggering conditions include a first triggering condition, a second triggering condition, and a third triggering condition; the first triggering condition is a preset triggering condition set based on a preset detection cycle; the second triggering condition is a preset triggering condition determined based on the real-time operating status of the target vehicle; and the third triggering condition is a preset triggering condition determined based on the historical anomaly detection results of the target vehicle.

3. The method for detecting anomalies in a vehicle-mounted camera according to claim 1, characterized in that, The step of analyzing the target statistical characteristics of the target media file to obtain the corresponding first analysis result includes: Based on the device type, installation environment, and historical operating data of the target camera, a target threshold is determined for the target camera; the target threshold includes a grayscale average threshold, a grayscale variance threshold, and a brightness variation threshold. Based on the target threshold, the target statistical characteristics of the target media file corresponding to the target camera are analyzed to obtain the corresponding first analysis result.

4. The method for detecting anomalies in a vehicle-mounted camera according to claim 1, characterized in that, Before performing reference region verification on the target media file based on the target reference region corresponding to the target camera, the method further includes: Determine the target reference area corresponding to the target camera based on the device type of the target camera; Wherein, if the target camera is an interior camera of the vehicle, the target reference area includes the steering wheel and the human face; if the target camera is an exterior camera of the vehicle, the target reference area includes the driving road and lane lines.

5. The method for detecting anomalies in a vehicle-mounted camera according to claim 4, characterized in that, The step of using the target detection model to perform reference region verification on the target media file based on the target reference region corresponding to the target camera, in order to obtain the corresponding second analysis result, includes: If the target camera is an interior camera used to capture images of the occupants of the target vehicle, then the first detection model is used to perform a first reference region verification on the target media file based on the target reference region; If the first reference region verification of the target media file is completed within the preset detection period, the corresponding second analysis result is obtained based on the first reference region verification result of the target media file. If the first reference region verification of the target media file is not completed within the preset detection period, the second detection model is used to perform a second reference region verification on the target media file based on the target reference region to obtain the corresponding second reference region verification result, and the corresponding second analysis result is obtained based on the second reference region verification result. The first detection model is a detection model that determines the human facial structure by recognizing key facial feature points, and the second detection model is a YOLO model used to recognize the human head; the detection efficiency of the first detection model is higher than that of the second detection model, and the detection accuracy of the second detection model is higher than that of the first detection model.

6. The method for detecting anomalies in a vehicle-mounted camera according to claim 1, characterized in that, After obtaining the anomaly detection result corresponding to the target camera, the method further includes: The anomaly detection results are saved to the target storage space, and all the anomaly detection results of the target vehicle within the preset time window are obtained from the target storage space based on the preset trend analysis period, so as to determine the target anomaly trend of the target camera of the target vehicle within the preset time window based on the anomaly detection results using preset trend analysis rules.

7. The method for detecting anomalies in a vehicle-mounted camera according to claim 6, characterized in that, Also includes: The preset triggering conditions are adjusted based on the target anomaly trend to obtain the adjusted preset triggering conditions, and then the process jumps to the step of sending an anomaly detection command to the target vehicle's infotainment system based on the preset triggering conditions.

8. An anomaly detection device for a vehicle-mounted camera, characterized in that, Applied to the target cloud, including: The media file acquisition module is used to send an anomaly detection command to the target vehicle's infotainment system based on preset trigger conditions, so as to acquire the target media file collected by the target vehicle's infotainment system using the target camera based on the anomaly detection command; the target media file includes an image to be detected and a video segment to be detected. The media file analysis module is used to analyze the target statistical characteristics of the target media file to obtain the corresponding first analysis result, and to determine the target detection model based on the target camera corresponding to the target media file. Using the target detection model, the target media file is used to perform reference region verification based on the target reference region corresponding to the target camera to obtain the corresponding second analysis result. The detection result acquisition module is used to determine the first confidence level corresponding to the first analysis result and the second confidence level corresponding to the second analysis result using a target confidence algorithm, and to fuse the first analysis result and the second analysis result based on the first confidence level and the second confidence level to obtain the anomaly detection result corresponding to the target camera.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the vehicle-mounted camera anomaly detection method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store a computer program, wherein the computer program, when executed by a processor, implements the vehicle camera anomaly detection method as described in any one of claims 1 to 7.

Images