Bicycle detection method, device, equipment and medium

Abstract

This application relates to a bicycle detection method, apparatus, device, and medium. The bicycle detection method comprises: identifying parked bicycles in a target parking area; obtaining target driving image data based on the parked bicycles; and performing image analysis based on the target driving image data to obtain brake detection results for the parked bicycles. This application utilizes image processing to detect bicycle brake anomalies, resolving the high installation and accessory costs associated with existing bicycle brake anomaly detection technologies, which require sensors to be installed on the bicycle handlebars. This reduces bicycle costs.

Description

Technical Field

[0001] The present application relates to the field of vehicle detection technology, and in particular to a single-vehicle detection method, device, equipment, and medium. Background Art

[0002] With the rapid development of mobile internet, shared bikes are becoming increasingly popular, bringing great convenience to people's travel. Given the large number of shared bikes, identifying bikes with abnormal brakes and promptly dispatching maintenance personnel for repairs has become an increasingly important issue in the shared bike industry.

[0003] Specifically, during the use of shared bicycles, due to long-term use, some brake components are severely worn, resulting in a decrease in braking performance, which poses a significant safety hazard to users. Currently, the commonly used brake anomaly detection method on the market mainly involves installing sensors on the bicycle handlebars. The information collected by the sensors determines the change in driving speed after the brakes are pressed, and determines whether the bicycle has a brake anomaly. However, this brake anomaly detection method requires the installation of sensors on the bicycle handlebars, which increases the additional installation and accessory costs of the bicycle, and cannot be applied to bicycles without sensors installed, resulting in a limited scope of application. Summary of the Invention

[0004] In view of this, in order to solve the above technical problems or part of the technical problems, the embodiments of the present application provide a bicycle detection method, device, equipment and medium.

[0005] In a first aspect, an embodiment of the present application provides a bicycle detection method, comprising:

[0006] Determine the parked bicycles in the target parking area;

[0007] Acquiring target driving image data based on the parked bicycle;

[0008] Image analysis is performed based on the target driving image data to obtain a brake detection result of the parked bicycle.

[0009] In a possible implementation, determining the parked bicycles in the target parking area includes:

[0010] Acquire image data captured by a camera in the target parking area;

[0011] Performing image recognition based on the captured image data to obtain recognized image information;

[0012] If the recognized image information includes target vehicle information, the target vehicle corresponding to the target vehicle information is determined to be the parked bicycle.

[0013] In a possible implementation, acquiring target driving image data based on the parked bicycle includes:

[0014] Extracting the pre-parking video corresponding to the parked bicycle from the surveillance video of the target parking area;

[0015] Frame processing is performed based on the pre-parking video to obtain the target driving image data.

[0016] In one possible implementation, the target driving image data includes at least two frames of bicycle image data in the pre-parking video, and performing image analysis based on the target driving image data to obtain the brake detection result of the parked bicycle includes:

[0017] Performing image recognition on the at least two frames of bicycle image data to obtain foot landing detection information and / or braking action detection information;

[0018] Analyze the bicycle position corresponding to each frame of bicycle image data to obtain braking speed detection information;

[0019] The braking detection result is determined based on the braking speed detection information, the foot landing detection information and / or the braking action detection information.

[0020] In a possible implementation, analyzing the bicycle position corresponding to each frame of bicycle image data to obtain braking speed detection information includes:

[0021] Determine the time information and bicycle position of each frame of bicycle image data;

[0022] Determine a bicycle distance difference between each two frames of bicycle image data according to the bicycle position, and determine a time difference between each two frames of bicycle image data according to the time information;

[0023] The braking speed detection information corresponding to the parked bicycle is determined based on the bicycle distance difference and the time difference.

[0024] In a possible implementation, determining the braking speed detection information corresponding to the parked bicycle based on the bicycle distance difference and the time difference includes:

[0025] determining a speed of the parked bicycle before parking based on the bicycle distance difference and the time difference;

[0026] determining braking change speed information of the parked bicycle according to the speed before parking;

[0027] If the braking change speed information belongs to braking deceleration information, the braking speed detection information is determined based on the deceleration index weight information corresponding to the single-vehicle distance difference.

[0028] In a possible implementation, performing image recognition on the at least two frames of bicycle image data to obtain foot landing detection information includes:

[0029] Identifying target frame image data from the at least two frames of bicycle image data, wherein a bicycle pedal position in the target frame image data is lower than a human foot position in the target frame image data;

[0030] Determining a landing parking distance based on the target frame image data;

[0031] If the landing distance is within the preset parking range, determining the foot landing index weight information corresponding to the parked bicycle according to the landing duration corresponding to the target frame image data;

[0032] The foot landing index weight information is determined as the foot landing detection information.

[0033] In a possible implementation, performing image recognition on the at least two frames of bicycle image data to obtain braking action detection information includes:

[0034] identifying braking action frame image data from the at least two frames of bicycle image data, wherein the braking action frame image data is bicycle image data including a palm feature object in a brake gripping state;

[0035] determining the number of braking actions according to time information corresponding to the braking action frame image data;

[0036] Braking action index weight information is determined according to the number of braking actions, and the braking action index weight information is determined as the braking action detection information.

[0037] In a possible implementation, determining the braking detection result based on the braking speed detection information, the foot landing detection information, and / or the braking action detection information includes:

[0038] determining a comprehensive brake abnormality index of the parked bicycle based on the brake speed detection information, the foot landing detection information, and the brake action detection information;

[0039] If the comprehensive braking abnormality index exceeds a preset abnormality index threshold, a braking abnormality detection result is sent according to the positioning position information of the parked bicycle, and the braking abnormality detection result is determined as the braking detection result.

[0040] In a second aspect, an embodiment of the present application provides a bicycle detection device, comprising:

[0041] A parked bicycle determination module is used to determine parked bicycles in a target parking area;

[0042] A target driving image module, configured to obtain target driving image data based on the parked bicycle;

[0043] The image analysis module is used to perform image analysis based on the target driving image data to obtain the brake detection result of the parked bicycle.

[0044] In a third aspect, an embodiment of the present application provides an electronic device comprising: a processor, and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the bicycle detection method as described in any one of the first aspects.

[0045] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the bicycle detection method as described in any one of the first aspects is implemented.

[0046] The bicycle detection method, device, equipment and medium provided in the embodiments of the present application determine the parked bicycles in the target parking area, and obtain target driving image data based on the parked bicycles, so as to perform image analysis based on the target driving image data to determine whether there is a risk of brake abnormality in the parked bicycle, thereby obtaining a brake detection result of the parked bicycle, achieving the purpose of using image processing to realize bicycle brake abnormality detection, solving the problem of high bicycle installation and accessories costs caused by the need to install sensors on the bicycle handlebars in existing bicycle brake abnormality detection technologies, and reducing bicycle costs.

[0047] In addition, the embodiment of the present application realizes bicycle brake abnormality detection based on image processing, without the need to install sensors on the bicycle handlebars, which solves the problem that the existing bicycle brake abnormality detection method requires the installation of sensors on the bicycle handlebars, resulting in a limited application range, and has a wide range of applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A flowchart of the steps of a bicycle detection method provided in an embodiment of the present application;

[0049] Figure 2 A flowchart of a bicycle detection method provided in an optional embodiment of the present application;

[0050] Figure 3 This is a schematic diagram of the arc between the tip of the index finger and the plane of the back of the hand in an example of this application;

[0051] Figure 4 This is a structural block diagram of a bicycle detection device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0052] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0053] To facilitate understanding of the embodiments of the present application, further explanation will be given below with reference to specific embodiments in conjunction with the accompanying drawings. The embodiments do not constitute a limitation on the embodiments of the present application.

[0054] Reference Figure 1 , shows a flowchart of the steps of a bicycle detection method provided by an embodiment of the present application. In a specific implementation, the bicycle detection method provided by an embodiment of the present application may include the following steps:

[0055] Step 110: Determine the parked bicycles in the target parking area.

[0056] Among them, the target parking area may refer to a preset bicycle parking area, such as a parking area pre-set for shared bicycles. Specifically, the embodiment of the present application can determine whether the target parking area has a bicycle to be detected parked therein by monitoring the target parking area. If the target parking area has a bicycle to be detected parked therein, the bicycle to be detected can be determined as a parked bicycle in the target parking area; if the target parking area does not have a bicycle to be detected, it can be determined that there is no bicycle parked in the target parking area. For example, monitoring equipment such as cameras and infrared sensors can be used to monitor whether the target parking area has a shared bicycle to be detected parked therein, so that when a shared bicycle to be detected is parked therein, the shared bicycle to be detected can be determined as a parked bicycle in the target parking area. It should be noted that in addition to shared bicycles, the bicycle to be detected can also be other pre-set types of bicycles, and the embodiment of the present application does not impose specific restrictions on this.

[0057] Step 120 : acquiring target driving image data based on the parked bicycle.

[0058] The target driving image data may include driving image data of a parked bicycle during a period of time before parking, and the driving image data can be used to represent a driving image before parking. Specifically, after determining a parked bicycle in a target parking area, the embodiment of the present application can retrieve a video of the parked bicycle before parking, and convert the video into driving image data by frame segmentation. The converted driving image data can then be used as the target driving image data, so that image analysis can be performed based on the target driving image data to determine whether the parked bicycle has a risk of abnormal braking.

[0059] Step 130 : performing image analysis based on the target driving image data to obtain a brake detection result of the parked bicycle.

[0060] Specifically, after acquiring the target image data, the embodiment of the present application can perform analysis and identification based on the target image data to determine whether there is a risk of abnormal braking of the parked bicycle, so as to generate a brake abnormality detection result when there is a risk of abnormal braking of the parked bicycle, as the brake detection result of the parked bicycle; and generate a normal brake detection result when there is no risk of abnormal braking of the parked bicycle, as the brake detection result of the parked bicycle, thereby achieving the purpose of realizing bicycle brake abnormality detection by using image processing.

[0061] It can be seen that the embodiment of the present application determines the parked bicycles in the target parking area, and obtains target driving image data based on the parked bicycles, and performs image analysis based on the target driving image data to determine whether there is a risk of abnormal braking of the parked bicycles, thereby obtaining the brake detection results of the parked bicycles, achieving the purpose of using image processing to realize bicycle brake abnormality detection, solving the problem of high bicycle installation and accessories costs caused by the need to install sensors on the bicycle handlebars in existing bicycle brake abnormality detection technologies, and reducing bicycle costs.

[0062] In actual processing, surveillance cameras within the parking area can be used to capture video footage of the bicycle before it stops, and image analysis can be performed to determine whether the user exhibited any abnormal braking action before parking. Optionally, based on the above-described embodiment, the present embodiment of the present invention can obtain target driving image data based on the parked bicycle, specifically by: extracting the corresponding pre-parking video footage of the parked bicycle from the surveillance video of the target parking area; and performing frame processing based on the pre-parking video to obtain the target driving image data. Subsequently, image recognition and analysis can be performed based on the target driving image data to determine whether the user exhibited any abnormal braking action before parking, i.e., detecting whether there was a corresponding abnormal braking action before the parked bicycle. Upon detecting abnormal braking action, it can be determined that the user's parked bicycle presents a risk of abnormal braking, and a corresponding brake detection result can be generated. This brake detection result can then be used to notify maintenance personnel to repair the bicycle, thereby resolving safety hazards associated with the user's use of the bicycle. The pre-parking video can refer to a video of the bicycle before it is parked; and the target driving image data can include one or more frames of bicycle image data extracted from the corresponding pre-parking video of the parked bicycle, though this embodiment does not impose specific limitations on this.

[0063] Reference Figure 2 , shows a flowchart of the steps of a bicycle detection method provided by an optional embodiment of the present application. Specifically, the bicycle detection method provided by the optional embodiment of the present application may include the following steps:

[0064] Step 210: Determine the parked bicycles in the target parking area.

[0065] In a specific implementation, one or more cameras may be pre-installed in the target parking area to perform image recognition by acquiring images captured by the cameras to determine whether a bicycle to be detected is parked in the target parking area. Thus, when a bicycle to be detected is parked in the target parking area, the bicycle to be detected can be determined as a parked bicycle in the target parking area. Furthermore, the embodiment of the present application determines the parked bicycle in the target parking area, which may specifically include: acquiring captured image data collected by a camera in the target parking area; performing image recognition based on the captured image data to obtain recognition image information; if the recognition image information contains target vehicle information, then determining the target vehicle corresponding to the target vehicle information as the parked bicycle. The captured image data may be used to represent an image captured by a camera, such as image data captured by a camera device, which may be a device provided with a camera; the target vehicle may refer to the bicycle to be detected, and the target vehicle information may be used to represent the bicycle to be detected.

[0066] Specifically, the embodiment of the present application can perform image recognition by acquiring the captured image data captured by the camera in the target parking area to obtain image recognition information. Subsequently, based on the image recognition information, it can be determined whether there is a bicycle to be detected parked in the target parking area. Thus, when there is a bicycle to be detected parked in the target parking area, the bicycle parked in the target parking area can be determined. For example, if the shared bicycle image information is set as the target vehicle information in advance, it can be determined whether there is a shared bicycle to be detected parked in the target parking area by determining whether the image recognition information contains the shared bicycle image information. If the image recognition information does not contain the shared bicycle image information, it can be determined that there is no shared bicycle to be detected parked in the target parking area, that is, there is no parked bicycle to be detected in the target parking area. If the image recognition information contains the shared bicycle image information, it can be determined that there is a shared bicycle to be detected parked in the target parking area. Then, the shared bicycle to be detected currently parked in the target parking area can be determined as a parked bicycle in the target parking area, and step 220 is performed for the parked bicycle in the target parking area.

[0067] Step 220: Extract the pre-parking video corresponding to the parked bicycle from the surveillance video of the target parking area.

[0068] Among them, the surveillance video may refer to a video recorded by a surveillance device such as a camera; the surveillance device can be used to monitor the target parking area and generate a corresponding recorded surveillance video. Specifically, in the embodiment of the present application, when a bicycle is determined to be parked in the target parking area, if there is a bicycle parked within the shooting field of view of a camera installed in the target parking area, the bicycle can be determined as a parked bicycle in the target parking area, and the surveillance video of the target parking area can be retrieved for the parked bicycle to extract the video of the bicycle before it is parked from the surveillance video, so that the extracted video can be determined as the pre-parking video corresponding to the parked bicycle, so that each frame of the pre-parking video can be tested later to determine whether there is a risk of abnormal braking of the parked bicycle.

[0069] Step 230 : performing frame processing based on the pre-parking video to obtain the target driving image data.

[0070] Step 240 : performing image analysis based on the target driving image data to obtain a brake detection result of the parked bicycle.

[0071] Specifically, after extracting the pre-parking video corresponding to the parked bicycle, the embodiments of the present application can perform frame processing on the pre-parking video by frame segmentation to convert the pre-parking video into driving image data. The converted driving image data can then be determined as target driving image data. Image analysis can then be performed based on the target driving image data to determine whether the user performed abnormal braking action before parking the bicycle. In other words, it can be detected whether the user of the parked bicycle performed abnormal braking action before parking. When abnormal braking action is detected, it can be determined that the parked bicycle is at risk of braking abnormality, and a corresponding brake abnormality detection result can be generated. This brake abnormality detection result can be used as the brake detection result of the parked bicycle, so that maintenance personnel can be notified based on the brake abnormality detection result to repair the parked bicycle at risk of braking abnormality. Abnormal braking action can include the following situations: the user's prolonged use of one or both feet before braking, the prolonged use of the vehicle handlebars while the vehicle is moving, and the vehicle deceleration before parking is too gentle, etc., which are not limited in the embodiments of the present application.

[0072] Optionally, in the embodiment of the present application, when the target driving image data includes at least two frames of bicycle image data in the pre-parking video, the image analysis based on the target driving image data to obtain the brake detection result of the parked bicycle may specifically include the following sub-steps:

[0073] Sub-step 2401 : performing image recognition on the at least two frames of bicycle image data to obtain foot landing detection information and / or braking action detection information.

[0074] Specifically, after converting the pre-parking video into target driving image data, the embodiments of the present application can perform image recognition on each frame of the target driving image data to determine whether the user's feet landed for a long time before braking. For example, image recognition of the person's feet in the parking video can be performed in a frame-by-frame manner to detect whether the person's feet are away from the bicycle pedals and close to the ground, and then determine whether one or both feet have landed, and generate corresponding foot landing detection information. This foot landing detection information can be used to determine the situation of the person's feet landing before the user brakes, such as the duration of the person's feet landing, the landing distance corresponding to the duration of the person's feet landing, and / or the weight information of the foot landing index, etc. The embodiments of the present application do not impose specific restrictions on this.

[0075] Furthermore, the embodiment of the present application performs image recognition on the at least two frames of bicycle image data to obtain foot landing detection information, which may specifically include: identifying target frame image data from the at least two frames of bicycle image data, the bicycle pedal position in the target frame image data being lower than the human foot position in the target frame image data; determining a landing parking distance based on the target frame image data; if the landing parking distance is within a preset parking range, determining the foot landing index weight information corresponding to the parked bicycle based on the landing duration corresponding to the target frame image data; and determining the foot landing index weight information as the foot landing detection information. Specifically, the embodiment of the present application can identify each frame of bicycle image data to identify the bicycle pedal position and the human foot position in each frame of bicycle image data, and can record the lowest position of the bicycle pedal, so as to determine whether the human foot is stepping on the pedal position by judging whether the bicycle pedal position in each frame of bicycle image data is directly above the lowest position of the bicycle pedal, and can judge whether the human foot position in the frame of bicycle image data is lower than the lowest position of the bicycle pedal to determine whether the foot has touched the ground; if the human foot is not stepping on the pedal position in each frame of bicycle image data (that is, the human foot is not directly above the pedal), and the human foot position is lower than the lowest position of the bicycle pedal, then it can be judged that the foot has touched the ground in this frame of bicycle image data, and then this frame of bicycle image data can be determined as the target frame image data. Subsequently, the landing distance can be determined based on the position of the bicycle in the target frame image data, and it can be determined whether the landing distance is within the preset parking range, so that the landing duration can be recorded when the landing distance is within the preset parking range to determine it as the landing duration corresponding to the target frame image data; and the target frame image data can be ignored when the landing distance is not within the preset parking range. After determining the landing duration corresponding to the target frame image data, the embodiment of the present application can determine the foot landing index weight information corresponding to the parked bicycle based on the landing duration corresponding to the target frame image data, and can determine the foot landing index weight information as the foot landing detection information, so that the foot landing detection information can be used to generate the brake detection result of the parked bicycle. Among them, the foot landing index weight information can refer to the weight value of the driving foot landing index. It should be noted that the longer the foot landing duration, the greater the weight value of the driving foot landing index.

[0076] In actual processing, in addition to determining whether there is a risk of abnormal braking of a parked bicycle by identifying from the bicycle image data whether the user has put his feet down for a long time before braking, the embodiments of the present application can also determine whether there is a risk of abnormal braking of a parked bicycle by other means. For example, the bicycle image data can be used to identify whether the vehicle handle is gripped with the brakes for a long time while the vehicle is moving, so as to determine the risk of abnormal braking when the vehicle handle is gripped with the brakes for a long time, etc. The embodiments of the present application do not limit this.

[0077] Furthermore, the embodiment of the present application performs image recognition on the at least two frames of bicycle image data to obtain braking action detection information, which may specifically include: identifying braking action frame image data from the at least two frames of bicycle image data, the braking action frame image data being bicycle image data containing a palm feature object in a brake gripping state; determining the number of braking actions based on the time information corresponding to the braking action frame image data; determining braking action index weight information based on the number of braking actions, and determining the braking action index weight information as the braking action detection information. Specifically, the embodiment of the present application can identify the bicycle handlebar position in each frame of bicycle image data to determine whether the bicycle handlebar is in a brake gripping state through image recognition, that is, to determine whether the handlebar has a brake gripping action. Specifically, if it is identified that the bicycle handlebars are in a brake gripping state in a certain frame of bicycle image data, it can be determined that the handlebars have a brake gripping action, and this frame of bicycle image data can be determined as brake action frame image data. The time information corresponding to the brake action frame image data can then be recorded to determine the number of brake actions based on the recorded time information, and thus the brake action index weight information can be determined based on the number of brake actions, so that the brake action index weight information can be used as brake action detection information to generate a brake detection result for a parked bicycle. Among them, the brake action index weight information can refer to the weight value of the action index of the bicycle handlebars having a brake gripping action for a long time. It should be noted that the more brake actions there are, that is, the more times the action of gripping the brake occurs, the greater the weight value of the brake gripping action index.

[0078] Sub-step 2402 , analyzing the bicycle position corresponding to each frame of bicycle image data to obtain braking speed detection information.

[0079] Specifically, when converting the pre-parking video into bicycle image data, the embodiment of the present application can record the time information of each frame of bicycle image data and mark the position of the bicycle in the frame of bicycle image data, so that the bicycle distance difference between every two frames (or every second) can be determined based on the recorded time information and the marked bicycle position. Then, by judging whether the bicycle distance difference gradually decreases over time, it can be determined that the driving speed of the parked bicycle has slowed down before parking. If the bicycle distance difference does not decrease significantly over time, it can be determined that the driving speed of the parked bicycle has not slowed down before parking, and then the corresponding braking speed detection information of the parked bicycle can be generated based on the bicycle distance difference between every two frames (or every second).

[0080] Furthermore, the embodiment of the present application performs analysis based on the bicycle position corresponding to each frame of bicycle image data to obtain brake speed detection information, which may include: determining the time information and bicycle position of each frame of bicycle image data; determining the bicycle distance difference between each two frames of bicycle image data based on the bicycle position, and determining the time difference between each two frames of bicycle image data based on the time information; and determining the brake speed detection information corresponding to the parked bicycle based on the bicycle distance difference and the time difference. Specifically, after determining the time information and bicycle position of each frame of bicycle image data, the embodiment of the present application can perform calculations based on the time information and bicycle position of each frame of bicycle image data to obtain the bicycle distance difference between each two frames of bicycle image data, and then can perform calculations based on the bicycle distance difference between each two frames of bicycle image data and the time difference between each two frames of bicycle image data to obtain the average speed corresponding to each two frames of bicycle image data. Subsequently, based on the average speed corresponding to each two frames of bicycle image data, it can be determined whether the deceleration of the parked bicycle before stopping is too gentle; if the deceleration of the parked bicycle before stopping is too gentle, it can be determined that an abnormal braking action has been detected, and further it can be determined that there is a risk of abnormal braking of the parked bicycle. Optionally, the embodiment of the present application determines the braking speed detection information corresponding to the parked bicycle based on the bicycle distance difference and the time difference, which can specifically include: determining the speed of the parked bicycle before stopping based on the bicycle distance difference and the time difference; determining the braking speed change information of the parked bicycle based on the speed before stopping; if the braking speed change information is braking deceleration information, the braking speed detection information can be determined based on the deceleration index weight information corresponding to the bicycle distance difference. Among them, the deceleration index weight information can refer to the weight value of the braking deceleration smoothness index. It should be noted that the smoother the braking deceleration, such as when the bicycle position distance difference is smaller, the greater the weight value of the braking deceleration smoothness index. If the brake change speed information does not belong to brake deceleration information, such as when the brake gripping action of the handlebar is detected, but the bicycle speed still does not change significantly or is in an accelerating state, it can be judged that the brake is abnormal, and the brake change speed information can be used as brake speed detection information, so that the brake detection result of the parked bicycle can be determined based on the brake speed detection information.

[0081] Sub-step 2403: determining the braking detection result based on the braking speed detection information, the foot landing detection information and / or the braking action detection information.

[0082] Specifically, after determining the brake speed detection information, the foot landing detection information and / or the brake action detection information, the embodiment of the present application can perform comprehensive calculation and processing based on the brake speed detection information, the foot landing detection information and / or the brake action detection information to obtain a brake abnormality comprehensive index, so that when the brake abnormality comprehensive index is greater than a preset brake abnormality index threshold, it can be determined that the parked bicycle is at risk of brake abnormality, and the corresponding brake detection result can be generated in combination with the geographical location of the parked bicycle, so that the corresponding maintenance personnel can be quickly matched to repair the parked bicycle based on the brake detection result to ensure the safety of the user using the bicycle. Further, the embodiment of the present application determines the brake detection result based on the brake speed detection information, the foot landing detection information and / or the brake action detection information, including: determining the brake abnormality comprehensive index of the parked bicycle based on the brake speed detection information, the foot landing detection information and the brake action detection information; if the brake abnormality comprehensive index exceeds the preset abnormality index threshold, the brake abnormality detection result is sent based on the positioning position information of the parked bicycle, and the brake abnormality detection result is determined as the brake detection result.

[0083] As an example of the present application, various human foot image data and pedal image data can be recorded through a preset deep learning system, and the recorded human foot image data and pedal image data can be used as a training set. Subsequently, the position of the human foot and the pedal in the image can be retrieved through machine learning, and the lowest position of the pedal can be recorded. It should be noted that when the pedal is below the wheel axis line, the farthest vertical distance between the pedal and the axis line is recorded as the lowest position of the pedal. If the human foot is not on the pedal position (i.e., the human foot is not directly above the pedal), and the human foot position is lower than the lowest position of the pedal, it can be determined that the foot has touched the ground. If the bicycle is parked more than 3 meters away from the user's parking position and the human foot is not on the pedal position, and this situation continues to occur within 1 meter of the parking position, it can be determined that the user's foot has touched the ground for a long time before braking, and the weight value P1 of the driving foot touchdown index V1 can be determined based on the foot touchdown duration, as the foot touchdown index weight information corresponding to the parked bicycle. It should be noted that, the longer the foot landing duration is, the greater the weight value P1 of the driving foot landing index V1 is.

[0084] In addition, this example can record various human hand image data while riding a bicycle through a deep learning system, and use the recorded human hand image data as a training set, and can retrieve the position of the human hand and index fingertip in the image through machine learning to obtain the curvature of the index fingertip and the back of the hand plane, so as to judge whether the palm object on the handlebar is in a clenched state through the curvature of the index fingertip and the back of the hand plane. Specifically, the back of the hand front data in the image data can be detected through image recognition, and then the curvature of the index fingertip and the back of the hand plane can be used to determine the condition of the palm object based on the back of the hand front data. If the fingertip or upper knuckle is detected based on the back of the hand front data, it can be judged that the palm object is in a semi-expanded state; if it is determined through image judgment that the hand is in contact with the brake handle, and the fingertip and upper knuckle cannot be detected based on the back of the hand front data, it can be judged that the palm is in a brake gripping state. Among them, the palm object can refer to the palm on the brake handle. For example, Figure 3 As shown, if the palm plane curvature a is between 0 and 60 degrees, the palm can be determined to be in a semi-extended state. If image recognition determines that the hand is in contact with the brake handle, and the palm plane curvature a is between 60 and 180 degrees, the palm can be determined to be in a clenched state. Through the above method, the transition of the palm object from a semi-extended state to a clenched state can be detected. If the transition from a semi-extended state to a clenched state is detected, it can be determined that a brake grip is present. If a brake grip occurs at a distance of more than 5 meters from the user's parking location and persists for more than 3 seconds, it is determined that the brake handle has been gripped for a long time while the vehicle was moving. Based on the number of brake grips, a weight value P2 for the brake grip action indicator V2 can be determined as brake action indicator weight information. It should be noted that the greater the number of brake grips, the greater the weight value P2 of the brake grip action indicator V2.

[0085] In actual processing, this example can convert the recorded pre-parking video into driving image data in a frame-by-frame manner as target driving image data, and record the system time of each target driving image data. Subsequently, a frame of target driving image data can be taken every 1S, and the position of the bicycle in the target driving image data in the frame can be marked as the bicycle position, so that the change in driving speed can be determined by obtaining the bicycle distance difference between every two frames (per second) to determine whether the driving speed has slowed down. If the bicycle distance difference does not become significantly smaller over time, it means that the driving speed has not slowed down. If the handle is detected to be gripped, but the speed still does not change significantly, it can be determined that the braking is abnormal. It should be noted that the smoother the driving deceleration, such as the bicycle position distance difference changes smaller and smaller over time, the greater the weight value P3 of the braking deceleration smoothness index V3. In addition, when calculating the vehicle speed, the vehicle's own positioning data can also be combined for reference, and the embodiment of the present application does not limit this.

[0086] It can be seen that this example can use the above method to determine whether a parked bicycle has abnormal braking action, and after determining that a parked bicycle has abnormal braking action, the braking abnormality comprehensive index can be obtained according to the weight value of each indicator. For example, it can be calculated according to the formula V = V1*P1+V2*P2+V3*P3 to obtain the braking abnormality comprehensive index of the parked bicycle. If the braking abnormality comprehensive index of the parked bicycle is greater than a certain threshold (i.e., the preset abnormality index threshold), it can be determined that the vehicle is at risk of braking abnormality, and the vehicle position can be located to generate the corresponding parking bicycle braking detection result in combination with the positioning position, so that the system can dispatch the nearest operation and maintenance personnel to handle it based on the braking detection result, saving manpower and financial costs.

[0087] In summary, the embodiments of the present application determine the parked bicycles in the target parking area, and obtain target driving image data based on the parked bicycles, and perform image analysis based on the target driving image data to determine whether the parked bicycles have the risk of braking abnormality, thereby obtaining the braking detection results of the parked bicycles, that is, bicycle braking abnormality detection is achieved based on image processing, without the need to install sensors on the bicycle handlebars, which solves the problem of high bicycle installation and accessories costs caused by the need to install sensors on the bicycle handlebars in existing bicycle brake abnormality detection technologies, reduces bicycle costs, and at the same time solves the problem of limited application scope caused by the need to install sensors on the bicycle handlebars in existing bicycle brake abnormality detection methods, and has a wide range of applications.

[0088] It should be noted that, for the purpose of simple description, the method embodiments are expressed as a series of action combinations, but those skilled in the art should know that the embodiments of the present application are not limited to the described order of actions, because according to the embodiments of the present application, certain steps can be performed in other orders or simultaneously.

[0089] The present application also provides a bicycle detection device. Figure 4 As shown, the bicycle detection device 300 may include the following modules:

[0090] a parked bicycle determination module 310 for determining parked bicycles in a target parking area;

[0091] A target driving image module 320 is configured to obtain target driving image data based on the parked bicycle;

[0092] The image analysis module 330 is used to perform image analysis based on the target driving image data to obtain the brake detection result of the parked bicycle.

[0093] Optionally, the parked bicycle determination module 310 may include the following submodules:

[0094] A captured image acquisition submodule is used to acquire captured image data collected by a camera in the target parking area;

[0095] An image recognition submodule, configured to perform image recognition based on the captured image data to obtain recognized image information;

[0096] The parked bicycle determination submodule is configured to determine, when the recognition image information includes target vehicle information, the target vehicle corresponding to the target vehicle information as the parked bicycle.

[0097] Optionally, the target driving image module 320 may include the following submodules:

[0098] a video extraction submodule, configured to extract the pre-parking video corresponding to the parked bicycle from the surveillance video of the target parking area;

[0099] The frame processing submodule is used to perform frame processing based on the pre-parking video to obtain the target driving image data.

[0100] Optionally, the target driving image data in the embodiment of the present application may include at least two frames of single-vehicle image data in the pre-parking video, and the image analysis module 330 may include the following submodules:

[0101] an identification submodule, configured to perform image recognition on the at least two frames of bicycle image data to obtain foot landing detection information and / or braking action detection information;

[0102] An analysis submodule is used to analyze the bicycle position corresponding to each frame of bicycle image data to obtain braking speed detection information;

[0103] The detection result determination submodule is used to determine the braking detection result based on the braking speed detection information, the foot landing detection information and / or the braking action detection information.

[0104] Optionally, the analysis submodule may include the following units:

[0105] A time and position determination unit, configured to determine the time information and position of each frame of bicycle image data;

[0106] a bicycle distance difference determining unit, configured to determine a bicycle distance difference between every two frames of bicycle image data according to the bicycle position;

[0107] a time difference determining unit, configured to determine a time difference between every two frames of bicycle image data based on the time information;

[0108] The braking speed detection unit is used to determine the braking speed detection information corresponding to the parked bicycle based on the bicycle distance difference and the time difference.

[0109] Optionally, the braking speed detection unit may include the following subunits:

[0110] a pre-parking speed subunit, configured to determine a pre-parking speed of the parked bicycle based on the bicycle distance difference and the time difference;

[0111] a brake change speed information subunit, configured to determine brake change speed information of the parked bicycle based on the pre-parking speed;

[0112] The braking speed detection information subunit is used to determine the braking speed detection information based on the deceleration index weight information corresponding to the single-vehicle distance difference when the braking change speed information belongs to braking deceleration information.

[0113] Optionally, the identification submodule may include the following units:

[0114] a target frame image recognition unit, configured to recognize target frame image data from the at least two frames of bicycle image data, wherein a bicycle pedal position in the target frame image data is lower than a human foot position in the target frame image data;

[0115] a landing and parking distance determining unit, configured to determine a landing and parking distance based on the target frame image data;

[0116] a first indicator weight information determining unit, configured to determine, when the foot landing distance is within a preset parking range, foot landing indicator weight information corresponding to the parked bicycle based on the foot landing duration corresponding to the target frame image data;

[0117] The foot landing detection information determining unit is used to determine the foot landing index weight information as the foot landing detection information.

[0118] Optionally, the identification submodule may include the following units:

[0119] a braking action frame recognition unit, configured to recognize braking action frame image data from the at least two frames of bicycle image data, wherein the braking action frame image data is bicycle image data containing a palm feature object in a brake gripping state;

[0120] a braking action number unit, configured to determine the number of braking actions based on time information corresponding to the braking action frame image data;

[0121] a second index weight information determining unit, configured to determine braking action index weight information according to the number of braking actions, and trigger a braking action detection information determining unit to determine the braking action index weight information as the braking action detection information;

[0122] The braking action detection information determining unit is configured to determine the braking action index weight information as the braking action detection information.

[0123] Optionally, the detection result determination submodule may include the following units:

[0124] a braking abnormality comprehensive index determining unit, configured to determine a braking abnormality comprehensive index of the parked bicycle based on the braking speed detection information, the foot landing detection information, and the braking action detection information;

[0125] a brake abnormality detection result sending unit, configured to send a brake abnormality detection result according to the positioning position information of the parked bicycle when the brake abnormality comprehensive index exceeds a preset abnormality index threshold;

[0126] The braking detection result determining unit is configured to determine the braking abnormality detection result as the braking detection result.

[0127] It should be noted that the bicycle detection device provided above can execute the bicycle detection method provided in any embodiment of the present application, and has the corresponding functions and beneficial effects of the execution method.

[0128] In a specific implementation, the above-mentioned bicycle detection device can be used in electronic devices, so that the electronic devices can use image processing to detect bicycle brake anomalies, solving the problem of high bicycle installation and accessories costs caused by the need to install sensors on the bicycle handlebars in existing bicycle brake anomaly detection technologies, reducing bicycle costs, and at the same time solving the problem of limited application scope caused by the need to install sensors on the bicycle handlebars in existing bicycle brake anomaly detection methods, thus having a wide range of applications. Furthermore, an embodiment of the present application also provides an electronic device, comprising: a processor, and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the bicycle detection method as described in any of the above-mentioned method embodiments.

[0129] An embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the bicycle detection method described in any of the above method embodiments are implemented.

[0130] It should be noted that, for the embodiments of the apparatus, device, and storage medium, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0131] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0132] Professionals should also be further aware that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians 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.

[0133] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of this application. It should be understood that the above description is only the specific implementation methods of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application should be included in the scope of protection of this application.

Claims

1. A bicycle detection method, characterized in that: include: Determine the parked bicycles in the target parking area; Acquiring target driving image data based on the parked bicycle; Performing image analysis based on the target driving image data to obtain a brake detection result of the parked bicycle; The target driving image data includes at least two frames of bicycle image data in the video before parking. The image analysis is performed based on the target driving image data to obtain the brake detection result of the parked bicycle, including: Performing image recognition on the at least two frames of bicycle image data to obtain foot landing detection information and / or braking action detection information; Analyze the bicycle position corresponding to each frame of bicycle image data to obtain braking speed detection information; The braking detection result is determined based on the braking speed detection information, the foot landing detection information and / or the braking action detection information.

2. The bicycle detection method according to claim 1, characterized in that: The step of determining the number of bicycles parked in the target parking area includes: Acquire image data captured by a camera in the target parking area; Performing image recognition based on the captured image data to obtain recognized image information; If the recognized image information includes target vehicle information, the target vehicle corresponding to the target vehicle information is determined to be the parked bicycle.

3. The bicycle detection method according to claim 1, characterized in that: Acquiring target driving image data based on the parked bicycle includes: Extracting the pre-parking video corresponding to the parked bicycle from the surveillance video of the target parking area; Frame processing is performed based on the pre-parking video to obtain the target driving image data.

4. The bicycle detection method according to any one of claims 1 to 3, characterized in that: The analysis of the bicycle position corresponding to each frame of bicycle image data to obtain the braking speed detection information includes: Determine the time information and bicycle position of each frame of bicycle image data; Determine a bicycle distance difference between each two frames of bicycle image data according to the bicycle position, and determine a time difference between each two frames of bicycle image data according to the time information; The braking speed detection information corresponding to the parked bicycle is determined based on the bicycle distance difference and the time difference.

5. The bicycle detection method according to claim 4, characterized in that: The determining, based on the bicycle distance difference and the time difference, the braking speed detection information corresponding to the parked bicycle includes: determining a speed of the parked bicycle before parking based on the bicycle distance difference and the time difference; determining braking change speed information of the parked bicycle according to the speed before parking; If the braking change speed information belongs to braking deceleration information, the braking speed detection information is determined based on the deceleration index weight information corresponding to the single-vehicle distance difference.

6. The bicycle detection method according to claim 1, characterized in that: The performing image recognition on the at least two frames of bicycle image data to obtain foot landing detection information includes: Identifying target frame image data from the at least two frames of bicycle image data, wherein a bicycle pedal position in the target frame image data is lower than a human foot position in the target frame image data; Determining a landing parking distance based on the target frame image data; If the landing distance is within the preset parking range, determining the foot landing index weight information corresponding to the parked bicycle according to the landing duration corresponding to the target frame image data; The foot landing index weight information is determined as the foot landing detection information.

7. The bicycle detection method according to claim 1, characterized in that: The performing image recognition on the at least two frames of bicycle image data to obtain braking action detection information includes: identifying braking action frame image data from the at least two frames of bicycle image data, wherein the braking action frame image data is bicycle image data including a palm feature object in a brake gripping state; determining the number of braking actions according to time information corresponding to the braking action frame image data; Braking action index weight information is determined according to the number of braking actions, and the braking action index weight information is determined as the braking action detection information.

8. The bicycle detection method according to claim 1, characterized in that: The determining of the braking detection result based on the braking speed detection information, the foot landing detection information and / or the braking action detection information includes: determining a comprehensive brake abnormality index of the parked bicycle based on the brake speed detection information, the foot landing detection information, and the brake action detection information; If the comprehensive braking abnormality index exceeds a preset abnormality index threshold, a braking abnormality detection result is sent according to the positioning position information of the parked bicycle, and the braking abnormality detection result is determined as the braking detection result.

9. A bicycle detection device, characterized in that: include: A parked bicycle determination module is used to determine parked bicycles in a target parking area; A target driving image module, configured to obtain target driving image data based on the parked bicycle; an image analysis module, configured to perform image analysis based on the target driving image data to obtain a brake detection result of the parked bicycle; The target driving image data includes at least two frames of bicycle image data in the pre-parking video, and the image analysis module is specifically configured to perform image recognition on the at least two frames of bicycle image data to obtain foot landing detection information and / or braking action detection information; and to obtain braking speed detection information based on the bicycle position corresponding to each frame of bicycle image data; The braking detection result is determined based on the braking speed detection information, the foot landing detection information and / or the braking action detection information.

10. An electronic device, characterized in that: include: A processor, and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the bicycle detection method as described in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the bicycle detection method according to any one of claims 1 to 8 is implemented.

Images