CN117173468A - Container door different opening detection method and device, electronic equipment and storage medium - Google Patents
Container door different opening detection method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN117173468A CN117173468A CN202311119908.XA CN202311119908A CN117173468A CN 117173468 A CN117173468 A CN 117173468A CN 202311119908 A CN202311119908 A CN 202311119908A CN 117173468 A CN117173468 A CN 117173468A
- Authority
- CN
- China
- Prior art keywords
- container
- container door
- opening detection
- target
- door
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 222
- 230000002159 abnormal effect Effects 0.000 claims abstract description 51
- 238000000034 method Methods 0.000 claims abstract description 40
- 238000012549 training Methods 0.000 claims abstract description 36
- 238000012545 processing Methods 0.000 claims abstract description 15
- 230000006870 function Effects 0.000 claims description 22
- 238000004590 computer program Methods 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 7
- 238000012544 monitoring process Methods 0.000 abstract description 8
- 238000004891 communication Methods 0.000 description 9
- 238000004364 calculation method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 239000003086 colorant Substances 0.000 description 2
- 238000012790 confirmation Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The application relates to the technical field of wharf port monitoring, and provides a container door different opening detection method, a device, electronic equipment and a storage medium, wherein the method comprises the following steps: acquiring image information of locks at two sides of a target container, wherein the locks comprise lock heads and lock rods; inputting the image information of the lockset into a container door different opening detection model to carry out container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the target container door, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container; based on the different opening detection result, whether the door of the target container is opened abnormally is determined. The container door abnormal opening detection model is obtained through training based on the Gaussian model of the image information structure of the lockset corresponding to the sample container, the lockset of the target container is automatically detected in real time, whether the container door is abnormally opened or not is judged, the monitoring efficiency of the container door is improved, and the safety performance is improved.
Description
Technical Field
The application relates to the technical field of wharf port monitoring, in particular to a container door different opening detection method, a device, electronic equipment and a storage medium.
Background
The container is used as a basic unit for port transportation, and can be frequently moved and moved during port container operation, so that the abnormal opening of the container door can be inevitably caused, and further, the cargo overflows, and safety accidents are caused. Although the equipment has a certain safety protection function, safety accidents are still very easy to occur due to the fact that the loading and unloading operation process is in a man-machine mixed mode and the environment is complex and changeable. Through the comprehensive application of modern sensing, detecting, identifying and controlling technology, the container loading and unloading efficiency and safety are improved, and the automation level of hoisting operation is improved, which is a trend in the industry.
However, the existing detection method for abnormal opening of the container door is usually monitored manually, and is low in efficiency and low in safety.
Disclosure of Invention
The present application aims to solve at least one of the technical problems existing in the prior art. Therefore, the application provides a container door abnormal opening detection method which is used for solving the defects of low monitoring efficiency and low safety of the existing container door abnormal opening detection method.
The application also provides a device for detecting the container door opening, electronic equipment and a storage medium.
According to an embodiment of the first aspect of the application, the method for detecting the container door ajar comprises the following steps:
acquiring image information of locks at two sides of a target container, wherein the locks comprise lock heads and lock rods;
inputting the image information of the lockset into a container door different opening detection model to perform container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container;
and determining whether the door of the target container is opened abnormally based on the different opening detection result.
According to the container door ajar detection method provided by the embodiment of the application, the method further comprises the step of training to obtain the container door ajar detection model:
constructing a preset target detection model;
acquiring image information of a lockset corresponding to the sample container and characteristic information of the lockset corresponding to the sample container, wherein the characteristic information represents parameter information of the lockset;
training the preset target detection model according to the image information of the lockset corresponding to the sample container and the characteristic information of the lockset corresponding to the sample container to obtain a target detection model and a corresponding detection result;
and carrying out Gaussian modeling on the detection result to obtain the container door different opening detection model.
According to the container door different opening detection method of the embodiment of the application, the training of the preset target detection model according to the image information of the lockset corresponding to the sample container and the characteristic information of the lockset corresponding to the sample container to obtain a target detection model and a corresponding detection result comprises the following steps:
inputting the image information of the lockset corresponding to the sample container into the preset target detection model for training to obtain a prediction detection result;
determining a target loss function value based on the difference between the prediction detection result and the characteristic information of the lockset corresponding to the sample container;
adjusting model parameters of the preset target detection model based on the target loss function value, and continuing iterative training based on the adjusted model parameters until reaching a preset training ending condition to obtain the target detection model and the corresponding detection result;
the preset training ending condition is that the target loss function value is smaller than a preset loss function value.
According to the container door ajar detection method of the embodiment of the application, the calculation formula of the target loss function value is as follows:
C＝γ 1 C cls +γ 2 C reg +γ 3 C cneter +γ 4 C h/w +γ 5 C angle
wherein, gamma 1-5 C is the corresponding weight value cls Representing class loss, C reg Rotation IOU penalty representing rotation rectangle, C cneter Representing center point information loss, C h/w Representing loss of aspect ratio information, C angle Indicating a loss of rectangular rotation angle information.
According to the container door open detection method provided by the embodiment of the application, the detection result comprises the category information, the rectangular center point information, the length and width information and the rectangular rotation angle information of the lockset, and the training result is subjected to Gaussian modeling to obtain the container door open detection model, and the method comprises the following steps:
and carrying out Gaussian modeling on the basis of the rectangular central point information, the length and width information and the rectangular rotation angle information of the lock rod of the lock head to obtain the container door different opening detection model.
According to the method for detecting the abnormal opening of the container door of the embodiment of the application, the method for determining whether the container door of the target container is abnormally opened according to the abnormal opening detection result comprises the following steps:
acquiring the different opening detection result corresponding to the image information of at least one lockset;
constructing a target detection result according to at least one different opening detection result;
judging whether the target detection result is smaller than a preset threshold value or not;
and when the target detection result is smaller than the preset threshold value, determining that the door of the target container is abnormally opened.
According to the container door open detection method provided by the embodiment of the application, before the image information of the lockset is input into the container door open detection model to carry out open detection processing, the method further comprises the following steps:
and correcting the distortion of the image information of the lockset based on a preset camera calibration method.
According to a second aspect of the present application, a container door ajar detection apparatus includes:
the acquisition module is used for acquiring the image information of locks at two sides of the target container, wherein the locks comprise lock heads and lock rods;
the detection module is used for inputting the image information of the lockset into a container door different opening detection model to carry out container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container;
and the determining module is used for determining whether the door of the target container is abnormally opened or not based on the different opening detection result.
An electronic device according to an embodiment of the third aspect of the present application includes a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the container door ajar detection method according to any of the above when executing the program.
A non-transitory computer readable storage medium according to an embodiment of the fourth aspect of the present application has stored thereon a computer program which, when executed by a processor, implements a container door ajar detection method as described in any of the above.
The above technical solutions in the embodiments of the present application have at least one of the following technical effects:
acquiring image information of locks at two sides of a target container, wherein the locks comprise lock heads and lock rods; inputting the image information of the lockset into a container door different opening detection model to perform container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container; and determining whether the door of the target container is opened abnormally based on the different opening detection result. The container door abnormal opening detection model is trained based on the Gaussian model of the lock corresponding to the sample container, the lock of the target container is automatically detected in real time, whether the container door is abnormally opened or not is judged, the monitoring efficiency of the container door is improved, the safety performance is improved, the port operation safety is guaranteed, and the economic loss caused by object falling and container collision is reduced.
Additional aspects and advantages of the application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the application.
Drawings
In order to more clearly illustrate the embodiments of the application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of a method for detecting a container door open in accordance with an embodiment of the present application;
FIG. 2 is a second flow chart of a method for detecting a container door opening in accordance with an embodiment of the present application;
FIG. 3 is a third flow chart of a method for detecting a container door opening in accordance with an embodiment of the present application;
fig. 4 is a schematic structural diagram of a container door differential opening detection device according to an embodiment of the present application;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present application more apparent, the technical solutions of the present application will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present application, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
In an optional embodiment, the camera a and the camera B mounted at the front end and the rear end of the container spreader acquire the image information of locks at two sides of the target container in real time, and the camera a and the camera B are vertically or obliquely downward overlook to capture the images of the door and the tail of the container; camera a and camera B may be connected to a computer that may acquire real-time images of camera a and camera B and process and detect them.
Optionally, the camera A and the camera B both adopt monitoring cameras, in order to meet the operation requirements of containers at ports and wharfs, the image resolution is 1980 multiplied by 1080 pixels, the frame rate is 30FPS, and the camera has a zooming function, and the zooming range is 6-135 mm. The camera A and the camera B are arranged at the head end and the tail end of the container lifting appliance through brackets, and the center of the camera after the installation is positioned at the middle position of the edge of the lifting appliance in the horizontal direction. Furthermore, the camera A and the camera B can observe most of the range of the front end and the rear end of the container by manually adjusting the angle and setting the focal length of the cradle head of the monitoring camera, and particularly, the two lock heads and the lock rod in the middle of the box door are required to be visible in the field of view of the camera image.
Optionally, the computer is connected with the camera A and the camera B through gigabit network cables, the computer can acquire real-time images of the two cameras, and the computer is provided with a high-performance central processing strong and graphic operation unit so as to monitor locks of container doors in real time.
Optionally, the computer a is in BUS communication with the container handling equipment control system PLC, preferably in the form of a pro fit-BUS, and specific communication is done in the form of TCP-IP. The PLC end of the tire crane control system is used as a service end, the computer A is used as a client end, the service end sends equipment working parameters such as lifting appliance height, lifting appliance opening and closing and the like to the client end in a certain period, the client end sends the current system health state to the service end in the same frequency, the current system health state comprises whether a camera is abnormal, whether a detection software system works normally and the like, and the service end can take data reported by the client end as heartbeat detection.
Optionally, the container handling equipment control system PLC controls the audible and visual alarm to generate an alarm. The equipment stops working, and meanwhile, the equipment control right is transferred to a manual operation platform, and the manual operation platform can receive alarm instructions and send alarm signals to workers through modules such as a display, a visual interface, a keyboard, a mouse, an indicator lamp, a loudspeaker and the like, and also can receive manual instructions such as braking, continuing, rechecking confirmation, rechecking non-confirmation and the like of the workers.
Fig. 1 is a schematic flow chart of a method for detecting a container door according to an embodiment of the present application, as shown in fig. 1, the method for detecting a container door includes:
step 101: acquiring image information of locks at two sides of a target container, wherein the locks comprise lock heads and lock rods;
in this embodiment, the cameras arranged at the two ends of the lifting appliance are used for acquiring the image information of the locks at the front and rear sides of the target container in real time, wherein the image information of the lock heads and the lock rods at the front and rear sides of the target container is particularly used for judging whether the container door is abnormally opened or not according to the image information of the lock heads and the lock rods.
Step 102: inputting the image information of the lockset into a container door different opening detection model to perform container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container;
it is easy to understand that, in this embodiment, before detecting the image information of the lock, the method further includes a step of training to obtain a container door differential opening detection model, when the container door is in a normal closed state, data of the position and the posture of the lock head and the lock rod are stored, and then modeling is performed according to the data, where the data modeling is performed by using a gaussian model method, and specifically, as shown in fig. 2, the method for training the container door differential opening detection model includes:
step 201: constructing a preset target detection model;
specifically, the preset target detection model is a rotating target detection algorithm based on a convolutional neural network, the convolutional neural network is composed of a main network, a multi-scale feature fusion network and a feature detection network, the input of the preset target detection model can be an RGB image with the size of 512 multiplied by 512 pixels, and the output is a three-channel output feature map, namely the output results are the types of a lock head and a lock rod, a rotating rectangular central point, a length and a width and a rectangular rotating angle respectively.
Step 202: acquiring image information of a lockset corresponding to the sample container and characteristic information of the lockset corresponding to the sample container, wherein the characteristic information represents parameter information of the lockset;
in one embodiment of the application, images of the doors of 200 different boxes in a normal closing state of the doors are collected in advance and are identified through a preset target detection model, at least 4 objects are identified in each image, namely a left lock and a lock rod, and a right lock and a lock rod, namely at least two pairs of lock locking rods are identified in each of the front side and the rear side of the container, the identified data format is (x, y, w, h, r), wherein x and y respectively represent the center point of a detection result, w and h represent the height and the width of the detection result, r represents the rotation angle of a rectangle, and a characteristic message represents the actual parameter information of the lockset.
Step 203: training the preset target detection model according to the image information of the lockset corresponding to the sample container and the characteristic information of the lockset corresponding to the sample container to obtain a target detection model and a corresponding detection result;
in this embodiment, as shown in fig. 3, training the preset target detection model according to the image information of the lock corresponding to the sample container and the feature information of the lock corresponding to the sample container, to obtain a target detection model and a corresponding detection result includes:
step 301: inputting the image information of the lockset corresponding to the sample container into the preset target detection model for training to obtain a prediction detection result;
step 302: determining a target loss function value based on the difference between the prediction detection result and the characteristic information of the lockset corresponding to the sample container;
in this embodiment, the latest method of deep learning target detection algorithm RTMdet (An Empirical Study of Designing Real-Time Object Detectors) is adopted. And optimizing its loss function as follows: when the cost of simOTA adopted in the original algorithm is calculated, the loss of the rectangular angle and the length-width ratio is added besides the center loss cost, and specifically, the calculation formula of the objective loss function value C is as follows:
C＝γ 1 C cls +γ 2 C reg +γ 3 C cneter +γ 4 C h/w +γ 5 C angle
wherein, gamma 1-5 C is the corresponding weight value cls Representing class loss, C reg Rotation IOU penalty representing rotation rectangle, C cneter Representing center point information loss, C h/w Representing loss of aspect ratio information, C angle Indicating a loss of rectangular rotation angle information.
Step 303: adjusting model parameters of the preset target detection model based on the target loss function value, and continuing iterative training based on the adjusted model parameters until reaching a preset training ending condition to obtain the target detection model and the corresponding detection result; the preset training ending condition is that the target loss function value is smaller than a preset loss function value.
In the training process of the model, regression and classification of the detection target frame are continuously carried out according to the current target loss function.
It can be understood that in the training process of the model, the lockset basically does not have larger scale change in combination with the use scene and working condition of the container, and more complex samples of colors, brightness, shadows and the like in image information caused by illumination change are adopted, so that the training intensity is improved by the original method of scaling, mosaic, mixUp and the like mixed pictures, the method of enhancing the data of the colors, brightness, shadows, translation, overturn and the like of the image is reserved, a large number of image training samples are manufactured, and the model is trained and verified by dividing the training set according to a preset proportion, so that the accuracy of model detection is improved. In an alternative embodiment, the sample is trained by making about 2 ten thousand images and following 2:8, training and verifying the model until the model reaches a preset training ending condition, for example, the preset training ending condition is that the calculated objective loss function value is smaller than the preset loss function value, or the preset training ending condition is that the objective loss function value obtains a minimum value or a minimum value, or the preset training ending condition is that the accuracy of model verification reaches a certain level, etc.
Step 204: and carrying out Gaussian modeling on the detection result to obtain the container door different opening detection model.
It is easy to understand that in this embodiment, the detection result includes the type information, the rectangular center point information, the length and width information, and the rectangular rotation angle information of the lock, and after the detection result is obtained, gaussian modeling is performed according to the data of the lock head and the lock rod that correspond to each other:
wherein mu is the expected value of the central point of the lock head or the angle of the two lock rods, sigma is the coordinate (x, y) of the central point of the lock head or the angle (theta 1 ，θ 2 ) Is a covariance matrix of (a).
It can be appreciated that in this embodiment, the mean and covariance matrix of the acquired two-dimensional data are calculated by first calculating the x-mean of the center coordinates or lock bar angles from the acquired dataAnd y is>As an unbiased estimate of μ, and considering that x has a correlation with y, the covariance matrix cov of x and y is calculated as an unbiased estimate of σ:
preferably, in order to make the model of this scheme succinct, only need carry out two-dimentional gaussian modeling to the central point information, the rectangular rotation angle information rectangle of length and width information and locking lever of tapered end, obtain container door and open the detection model, can satisfy and carry out abundant judgement to the state of tapered end and locking lever to detect whether container door is in the unusual open state.
It can be understood that in the stage of identifying and detecting the image information of the lock, the container door differential opening detection model in the above embodiment is adopted to perform reasoning identification, and possible false identification results, such as the relative position and the posture of the lock rod and the lock head, are taken out a priori. In the recognition and detection stage of the model, due to insufficient generalization performance of the model, error detection is easy to generate to a certain extent in the detection process, false alarm is further caused, and in order to ensure the accuracy of a detection result, the detection result can be removed through rotation of a non-maximum suppression (NMS) and a threshold value of a rectangle, and then further denoising is performed according to a Gaussian model corresponding to the length and width information of the lock head, so that an accurate detection result is obtained.
In an alternative embodiment, whether the current image is the door position is determined by detecting whether the lock and the lock bar are present in the image information and whether a corresponding positional relationship exists between the lock and the lock bar, such as the lock being above the lock bar. After the image corresponding to the box door is determined, in view of the fact that the camera in the scheme is mounted on the lifting appliance of the tire crane or the track crane, when the lifting appliance picks up the container, the camera and the container cannot generate relative movement, so that the box door can be determined by identifying the position, the gesture and other characteristics of the lock head and the lock rod, and whether the box door is abnormally opened or not is further determined.
Step 103: and determining whether the door of the target container is opened abnormally based on the different opening detection result.
In this embodiment, the determination may be performed according to the projection image of the gaussian model established as described above, and in practice, it is required to perform classification detection on the new data according to a value of a certain contour line of the projection image of the gaussian model corresponding to a threshold, that is, determine whether the new data can satisfy the current gaussian distribution. In the scheme, the mode of selecting the optimal threshold value Epsilon by adopting the maximum F1 score (F1-score) is adopted, namely, the calculation of the F1-score is carried out by collecting positive and negative samples, when the threshold value is maximum, the optimal probability density value Epsilon is selected as a judgment basis, the optimal probability density value Epsilon can be manually preset and adjusted, if the probability density value obtained after the calculation of new data is smaller than Epsilon, F=0 is judged, namely, the data is judged to be outlier or abnormal data, and otherwise F=1 is judged.
Further, in order to ensure the accuracy of the detection result, when identifying whether the door of the container is abnormally opened or not, that is, when identifying and detecting the image information of the lock, a multi-frame image data combination mode may be adopted to determine, for example, the detection result corresponding to the image information of the 3-frame lock is collected in a certain preset time, and the average value of the detection result is taken, and the determination is performed according to the determination mode, where the determination mode is as follows:
T＝α 1 L 1 +α 2 L 2 +α 3 L 3
in the above formula, alpha 1 、α 2 、α 3 For the weight value of each judgment condition, α is calculated for convenience in this embodiment 1 、α 2 、α 3 Are all set to 0.3, L 1 And L 2 Representing the judgment conditions of the left lock and the right lock respectively, F represents the judgment of the point coincidence model, V represents the judgment of the angle coincidence threshold, L depends on the judgment of the center point model and the judgment of the rectangular rotation angle 3 The judgment of the rectangular rotation angle model of the lock rod is only dependent on the judgment of the angle of the lock rod.
In this embodiment, a target detection result T is constructed by acquiring a different opening detection result corresponding to the image information of at least one lock, and according to the at least one different opening detection result, then whether the target detection result T is smaller than a preset threshold value is determined; and when the calculated target detection result T is smaller than a preset threshold value, determining that the door of the target container is abnormally opened. In the actual determination, the preset threshold may be set to 0.6, and when the calculated target detection result T is greater than or equal to 0.6, it is indicated that the container door is normally closed, otherwise, it is determined that the container door is abnormally opened.
Further, because the distance between the lenses of the camera a and the camera B and the lock of the container door is short, and the camera with the wide-angle lens is adopted in the scheme, barrel distortion is easy to generate in the shot image, before the image information of the lock is input into the container door different opening detection model for different opening detection, the distortion correction is firstly required to be performed on the shot image of the camera according to a preset camera calibration method, and optionally, the distortion correction is performed on the shot image by adopting a traditional Zhang Zhengyou camera calibration method.
It can be understood that after the container door is detected by the container door abnormal opening detection method, if the container door is detected to be abnormally opened, the detection result is reported by an event through a communication module of a logic controller PLC for processing the strong tyre crane or the track crane, the PLC controls the audible and visual alarm to give an alarm, and simultaneously controls the vehicle to stop working, stops lifting the container, and improves the safety performance in the working process.
According to the container door different opening detection method, image information of locks at two sides of a target container is obtained, wherein the locks comprise lock heads and lock rods; inputting the image information of the lockset into a container door different opening detection model to carry out container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container; based on the different opening detection result, whether the door of the target container is opened abnormally is determined. The container door abnormal opening detection model is trained based on the Gaussian model of the lock corresponding to the sample container, the lock of the target container is automatically detected in real time, whether the container door is abnormally opened or not is judged, the monitoring efficiency of the container door is improved, the safety performance is improved, the port operation safety is guaranteed, and the economic loss caused by object falling and container collision is reduced.
The container door opening detection device provided by the application is described below, and the container door opening detection device described below and the container door opening detection method described above can be correspondingly referred to each other.
Fig. 4 is a schematic structural diagram of a container door ajar detection device according to an embodiment of the present application, as shown in fig. 4, including:
an acquisition module 401, configured to acquire image information of locks on both sides of a target container, where the locks include a lock and a lock rod;
the detection module 402 is configured to input image information of the lock into a container door differential opening detection model to perform container door differential opening detection processing, so as to obtain a differential opening detection result; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container;
a determining module 403, configured to determine whether a door of the target container is opened abnormally based on the different opening detection result.
Based on any of the above embodiments, the detection module 402 includes:
constructing a preset target detection model;
acquiring image information of a lockset corresponding to the sample container and characteristic information of the lockset corresponding to the sample container, wherein the characteristic information represents parameter information of the lockset;
training the preset target detection model according to the image information of the lockset corresponding to the sample container and the characteristic information of the lockset corresponding to the sample container to obtain a target detection model and a corresponding detection result;
and carrying out Gaussian modeling on the detection result to obtain the container door different opening detection model.
Fig. 5 illustrates a physical schematic diagram of an electronic device, as shown in fig. 5, which may include: processor 510, communication interface (Communications Interface) 520, memory 530, and communication bus 540, wherein processor 510, communication interface 520, memory 530 complete communication with each other through communication bus 540. The processor 510 may invoke logic instructions in the memory 530 to perform a container door ajar detection method comprising: acquiring image information of locks at two sides of a target container, wherein the locks comprise lock heads and lock rods; inputting the image information of the lockset into a container door different opening detection model to perform container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container; and determining whether the door of the target container is opened abnormally based on the different opening detection result.
Further, the logic instructions in the memory 530 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present application also provides a computer program product, the computer program product comprising a computer program, the computer program being storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, being capable of executing the container door ajar detection method provided by the above methods, the method comprising: acquiring image information of locks at two sides of a target container, wherein the locks comprise lock heads and lock rods; inputting the image information of the lockset into a container door different opening detection model to perform container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container; and determining whether the door of the target container is opened abnormally based on the different opening detection result.
In yet another aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program which when executed by a processor is implemented to perform the container door ajar detection method provided by the above methods, the method comprising: acquiring image information of locks at two sides of a target container, wherein the locks comprise lock heads and lock rods; inputting the image information of the lockset into a container door different opening detection model to perform container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container; and determining whether the door of the target container is opened abnormally based on the different opening detection result.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present application without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application.
The above embodiments are only for illustrating the present application, and are not limiting of the present application. Although the present application has been described in detail with reference to the embodiments, it will be understood by those skilled in the art that various combinations, modifications, or equivalents may be made to the technical solutions of the present application without departing from the spirit and scope of the technical solutions of the present application, and the present application is intended to be covered in the protection scope of the present application.
Claims (10)
1. A container door ajar detection method, comprising:
acquiring image information of locks at two sides of a target container, wherein the locks comprise lock heads and lock rods;
inputting the image information of the lockset into a container door different opening detection model to perform container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container;
and determining whether the door of the target container is opened abnormally based on the different opening detection result.
2. The container door ajar detection method of claim 1 further comprising the step of training to obtain the container door ajar detection model:
constructing a preset target detection model;
acquiring image information of a lockset corresponding to the sample container and characteristic information of the lockset corresponding to the sample container, wherein the characteristic information represents parameter information of the lockset;
training the preset target detection model according to the image information of the lockset corresponding to the sample container and the characteristic information of the lockset corresponding to the sample container to obtain a target detection model and a corresponding detection result;
and carrying out Gaussian modeling on the detection result to obtain the container door different opening detection model.
3. The method for detecting the abnormal opening of the container door according to claim 2, wherein the training the preset target detection model according to the image information of the lock corresponding to the sample container and the characteristic information of the lock corresponding to the sample container to obtain the target detection model and the corresponding detection result comprises the following steps:
inputting the image information of the lockset corresponding to the sample container into the preset target detection model for training to obtain a prediction detection result;
determining a target loss function value based on the difference between the prediction detection result and the characteristic information of the lockset corresponding to the sample container;
adjusting model parameters of the preset target detection model based on the target loss function value, and continuing iterative training based on the adjusted model parameters until reaching a preset training ending condition to obtain the target detection model and the corresponding detection result;
the preset training ending condition is that the target loss function value is smaller than a preset loss function value.
4. A container door ajar detection method according to claim 3, wherein the objective loss function value is calculated by the formula:
C＝γ 1 C cls +γ 2 C reg +γ 3 C cneter +γ 4 C h/w +γ 5 C angle
wherein, gamma 1-5 C is the corresponding weight value cls Representing class loss, C reg Rotation IOU penalty representing rotation rectangle, C cneter Representing center point information loss, C h/w Representing loss of aspect ratio information, C angle Indicating a loss of rectangular rotation angle information.
5. The method for detecting the abnormal opening of the container door according to claim 3, wherein the detection result comprises category information, rectangular center point information, length and width information and rectangular rotation angle information of the lock, the training result is subjected to gaussian modeling to obtain the abnormal opening detection model of the container door, and the method comprises the following steps:
and carrying out Gaussian modeling on the basis of the rectangular central point information, the length and width information and the rectangular rotation angle information of the lock rod of the lock head to obtain the container door different opening detection model.
6. The container door ajar detection method of claim 1, wherein the determining whether the container door of the target container is abnormally opened based on the ajar detection result comprises:
acquiring the different opening detection result corresponding to the image information of at least one lockset;
constructing a target detection result according to at least one different opening detection result;
judging whether the target detection result is smaller than a preset threshold value or not;
and when the target detection result is smaller than the preset threshold value, determining that the door of the target container is abnormally opened.
7. The method for detecting the opening of the container door according to any one of claims 1 to 6, wherein before the image information of the lock is input into a container door opening detection model to perform the opening detection process, the method further comprises:
and correcting the distortion of the image information of the lockset based on a preset camera calibration method.
8. A container door ajar detection device, comprising:
the acquisition module is used for acquiring the image information of locks at two sides of the target container, wherein the locks comprise lock heads and lock rods;
the detection module is used for inputting the image information of the lockset into a container door different opening detection model to carry out container door different opening detection processing, and obtaining different opening detection results; the abnormal opening detection result represents the abnormal opening probability of the container door of the target container, and the container door abnormal opening detection model is a Gaussian model constructed according to the image information of the lockset corresponding to the sample container;
and the determining module is used for determining whether the door of the target container is abnormally opened or not based on the different opening detection result.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the container door ajar detection method of any one of claims 1 to 7 when the program is executed.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the container door ajar detection method of any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311119908.XA CN117173468A (en) | 2023-08-31 | 2023-08-31 | Container door different opening detection method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311119908.XA CN117173468A (en) | 2023-08-31 | 2023-08-31 | Container door different opening detection method and device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117173468A true CN117173468A (en) | 2023-12-05 |
Family
ID=88944284
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311119908.XA Pending CN117173468A (en) | 2023-08-31 | 2023-08-31 | Container door different opening detection method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117173468A (en) |
-
2023
- 2023-08-31 CN CN202311119908.XA patent/CN117173468A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107133973B (en) | Ship detection method in bridge collision avoidance system | |
| CN106056079B (en) | A kind of occlusion detection method of image capture device and human face five-sense-organ | |
| CN104573659B (en) | A kind of driver based on svm takes phone-monitoring method | |
| CN103984961A (en) | Image detection method for detecting foreign matter at bottom of vehicle | |
| CN104966304A (en) | Kalman filtering and nonparametric background model-based multi-target detection tracking method | |
| US20170278234A1 (en) | Method for detecting a defect on a surface of a tire | |
| CN107194393A (en) | A kind of method and device for detecting Provisional Number Plate | |
| CN111460917B (en) | Airport abnormal behavior detection system and method based on multi-mode information fusion | |
| CN115171218A (en) | Material sample feeding abnormal behavior recognition system based on image recognition technology | |
| CN113034378A (en) | Method for distinguishing electric automobile from fuel automobile | |
| CN115841633A (en) | Power tower and power line associated correction power tower and power line detection method | |
| CN113065454B (en) | High-altitude parabolic target identification and comparison method and device | |
| JP2021128796A (en) | Object recognition system and object recognition method | |
| CN111402185B (en) | Image detection method and device | |
| CN117173468A (en) | Container door different opening detection method and device, electronic equipment and storage medium | |
| CN110634120B (en) | Vehicle damage judgment method and device | |
| CN113221603A (en) | Method and device for detecting shielding of monitoring equipment by foreign matters | |
| CN112801072B (en) | Elevator non-flat-layer door opening fault recognition device and method based on computer vision | |
| CN115423764A (en) | Method and device for detecting shielding of camera and computer equipment | |
| CN108960181A (en) | Black smoke vehicle detection method based on multiple dimensioned piecemeal LBP and Hidden Markov Model | |
| CN115082850A (en) | Template support safety risk identification method based on computer vision | |
| CN116681955B (en) | Method and computing device for identifying traffic guardrail anomalies | |
| CN115713758B (en) | Carriage identification method, system, device and storage medium | |
| CN116740111A (en) | Forward-moving rotary avoidance type three-dimensional parking obstacle avoidance method and system based on machine vision | |
| CN117197710A (en) | Detection device and method for lock pad of container, electronic equipment and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |