CN115965625A - Instrument detection device based on visual identification and detection method thereof - Google Patents
Instrument detection device based on visual identification and detection method thereof Download PDFInfo
- Publication number
- CN115965625A CN115965625A CN202310252271.5A CN202310252271A CN115965625A CN 115965625 A CN115965625 A CN 115965625A CN 202310252271 A CN202310252271 A CN 202310252271A CN 115965625 A CN115965625 A CN 115965625A
- Authority
- CN
- China
- Prior art keywords
- instruction
- image
- detection
- acquiring
- instrument equipment
- 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.)
- Granted
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The invention belongs to the technical field of instrument detection, and particularly relates to an instrument detection device based on visual identification and a detection method thereof, wherein the detection device comprises the following steps: a detection model and an instruction library are established in a terminal server, wherein the instruction library comprises a plurality of instruction subsets; acquiring an image of instrument equipment through an image acquisition module, acquiring a panel image, identifying the panel image through the detection model, acquiring panel parameters, and acquiring a to-be-tested instruction set according to an initial characteristic matching instruction base of the panel; and selecting any unselected instruction subset from the instruction set to be tested, sending the instruction subset to instrument equipment, carrying out image acquisition on the instrument equipment to obtain a feedback atlas, and simultaneously interpreting the instruction subset through the terminal server to obtain instruction content. The invention identifies the panel characteristics by detecting the model and calls the corresponding test message to detect the instrument equipment, thereby improving the generalization capability and the application range of the model.
Description
Technical Field
The invention belongs to the technical field of instrument detection, and particularly relates to an instrument detection device based on visual identification and a detection method thereof.
Background
In recent years, with the rapid development of industrial automation technology, more and more enterprises and factories begin to adopt automated detection technology to improve product quality and production efficiency. Among them, the instrument visual recognition technology has been widely used and studied as one of the important means for automatic detection. The instrument visual recognition technology utilizes equipment such as a computer, a camera and the like to collect an instrument image, and realizes automatic detection and recognition of the instrument through methods such as image processing, feature extraction, classification and the like. The technology has the advantages of non-contact, high efficiency, high precision and the like, and is widely applied to industries such as manufacturing industry, electric power, energy, traffic and the like.
In the vehicle instrument equipment production industry, in the conventional technology, after the vehicle instrument equipment is produced, the icon on the vehicle instrument equipment is detected manually, the problems of low efficiency, insufficient accuracy and the like exist, the conditions of omission, false detection and the like also exist, the detection efficiency is improved along with the application of the visual identification technology to the vehicle instrument equipment, the qualification rate of products is improved, however, the visual identification technology is utilized to detect the vehicle instrument equipment, the application scenes of the vehicle instrument equipment are different, the situations of different shapes, different functions, different icon forms, different icon distribution positions and the like exist, different detection models are often required to be designed for different instrument equipment, and the problems of high design difficulty, poor model generalization capability and the like exist in the existing machine visual identification technology.
In addition, based on the conventional instrument detection mode, the potential hazards in the production process of the instrument cannot be determined according to the attributes of the instrument, for example, in a certain period of time, although the qualified rate of the instrument is higher than the standard qualified rate, problems cannot be displayed timely due to faults of production equipment, and then a large-scale rework situation may occur in the following process after a certain period of time.
Disclosure of Invention
The invention aims to provide an instrument detection method based on visual identification, which identifies the characteristics of a panel by detecting a model and calls a corresponding test message to detect instrument equipment, thereby improving the generalization capability and the application range of the model.
The technical scheme adopted by the invention is as follows:
a meter detection method based on visual identification mainly comprises the following steps:
s1: constructing a detection model and an instruction library on a terminal server, wherein the instruction library comprises a plurality of instruction subsets;
s2: acquiring an image of a panel by an image acquisition module, identifying the image of the panel by the detection model to acquire panel parameters, and matching an instruction library according to the panel parameters to acquire an instruction set to be tested;
s3: selecting any unselected instruction subset from the instruction set to be tested, sending the instruction subset to instrument equipment, carrying out image acquisition on the instrument equipment to obtain a feedback image set, and simultaneously interpreting the instruction subset through the terminal server to obtain detection contents;
s4: identifying a feedback atlas by using the panel image as a reference through a detection model to obtain difference characteristics, and matching the difference characteristics with detection contents to obtain a judgment record;
s5: and repeating the steps in the S3 and the S4 until all the instruction subsets in the instruction set to be tested are sent to the instrument equipment, acquiring a plurality of judgment records, generating a detection record according to the judgment records, and storing the detection record.
In a preferred embodiment, as in S1, the main steps of constructing the detection model at the terminal server are as follows:
s11: constructing a convolutional neural network model, wherein the convolutional neural network model at least comprises an input layer, a convolutional layer, a pooling layer, a full-link layer and an output layer;
s12: acquiring training data, verification data and test data, wherein the training data at least comprises a symbol identification atlas, a segment type icon atlas, a needle type atlas and a text type atlas;
s13: inputting the training data into a convolutional neural network for training to obtain a detection model based on the convolutional neural network;
s14: and verifying the detection model by using the verification data, and testing the detection model by using the test data.
In a preferred embodiment, as described in S1, the main steps of constructing the instruction library at the terminal server are as follows:
s15: constructing an instruction library, constructing a plurality of instruction subsets in the instruction library, and creating a primary signal tag for each instruction subset, wherein detection target information is recorded in the primary signal tag;
s16: and constructing a plurality of test messages in the instruction subset according to the primary signal tags, and creating a secondary signal tag for each test message, wherein the secondary signal tag records target state information.
In a preferred embodiment, as in S2, the main steps of obtaining the instruction set to be tested are as follows:
s21: acquiring an image of instrument equipment through an image acquisition module to obtain a panel image;
s22: identifying the panel image through the detection model, acquiring panel parameters containing a plurality of panel characteristics, and creating an image label for each panel characteristic, wherein panel characteristic information is recorded in the image label;
s23: and calling the instruction subset matched with the image tag and the primary signal tag according to the image tag matching instruction library to obtain an instruction set to be tested.
In a preferred embodiment, as in S3, the main steps of acquiring an image of the meter device and obtaining a feedback atlas are as follows:
s31: selecting any unselected instruction subset from the instruction set to be tested, and sending a plurality of test messages in the instruction subset to instrument equipment one by one;
s32: and when one test message is sent, the image acquisition module acquires an image of the instrument equipment once to acquire a feedback image, and acquires a feedback image set until all the test messages in the instruction subset are sent.
In a preferred embodiment, as in S4, the main steps of obtaining the determination record are as follows:
s41: inputting the feedback image and the panel image into a detection model, and acquiring the difference characteristics of the feedback image and the panel image;
s42: matching the difference characteristics with the instruction content to obtain a judgment record, wherein the judgment record comprises a feedback image, a test message, a secondary signal label and a judgment result;
s43: if the difference characteristics are not matched with the message content, judging that the result is unqualified;
s44: and if the difference characteristics are matched with the message content, judging that the result is qualified.
In a preferred embodiment, as described in S5, the step after generating a detection record according to a plurality of the determination records and storing the detection record further includes:
s51: counting the qualification rate of the instrument equipment in unit time;
s52: acquiring the qualification rate in all detection periods, and inputting the qualification rate into a trend change function to obtain a change trend value of the qualification rate of the instrument equipment; a trend change function of the yieldThe number is as follows:
in the formula (I), wherein,
represents a change trend value>
Representing the number of detection periods>
Represents a section 2 to>
The qualification rate in (4), based on the status of the status quo>
Represents a section 1 to>
-a yield in 1;
s53: obtaining an allowable deviation interval and comparing the allowable deviation interval with the variation trend value;
s54: if the variation trend value is within the allowable deviation interval, judging that the production line of the instrument equipment normally operates;
s55: and if the variation trend value is lower than the lower limit value of the allowable deviation interval, judging that the production line of the instrument equipment is abnormal in operation, and sending an alarm signal.
In a preferred embodiment, the step after obtaining the value of the trend of change of the yield of the meter device further includes:
s56: acquiring a yield change value in adjacent detection periods in real time;
s57: acquiring a fluctuation threshold value, and comparing the fluctuation threshold value with the qualified rate change value;
s58: if the qualified rate change value is higher than or equal to the fluctuation threshold value, judging that the production line of the instrument equipment is abnormal and not adding the abnormal production line into the trend change function;
s59: and if the qualified rate change value is lower than a fluctuation threshold value, adding the qualified rate change value into a trend change function.
The instrument detection device based on visual recognition is applied to the instrument detection method based on visual recognition, and comprises the following steps:
the terminal server is used for constructing a detection model and an instruction library;
the image acquisition module is used for acquiring images of the instrument equipment;
the judging module is used for generating a judging record according to the instruction content and the feedback atlas;
the alarm module is used for sending out an alarm signal;
an interpretation module capable of interpreting the test message.
A server device comprising a processor, a storage element, and a computer program stored on the storage element and executable by the processor, the processor implementing the steps of one of the visual recognition based meter detection methods described above when executing the computer program.
The invention has the technical effects that:
the method identifies the panel characteristics of the instrument equipment through the detection model, calls corresponding test messages according to the panel characteristics, sends the called test messages to the instrument equipment one by one, and identifies the panel characteristics sending feedback signals by matching with a machine identification technology so as to judge whether the instrument equipment is qualified or not, thereby improving the generalization capability and the application range of the model;
according to the invention, the qualified rate of the instrument equipment in a plurality of detection periods is collected to generate the qualified rate change trend value, the change trend of the qualified rate can be intuitively and conveniently known, whether the production line of the instrument equipment needs to be checked is judged according to the change trend of the qualified rate, so that problems can be found and solved in time, and the problem exposure of the production line is late, so that the yield of subsequent products has large-scale deviation.
Drawings
FIG. 1 is a schematic flow diagram of the present invention as a whole;
fig. 2 is a block diagram of the configuration of the server apparatus of the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention more comprehensible, embodiments accompanying figures of the present invention are described in detail below.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced otherwise than as specifically described herein, and it will be appreciated by those skilled in the art that the present invention may be practiced without departing from the spirit and scope of the present invention and that the present invention is not limited by the specific embodiments disclosed below.
Furthermore, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one implementation of the invention. The appearances of the phrase "in one preferred embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.
Referring to fig. 1, the present invention provides a method for detecting a meter based on visual recognition, which mainly comprises the following steps:
step1: constructing a detection model and an instruction library on a terminal server, wherein the instruction library comprises a plurality of instruction subsets;
step2: acquiring an image of instrument equipment through an image acquisition module, acquiring a panel image, identifying the panel image through the detection model, acquiring panel parameters, matching an instruction base according to the panel parameters, and acquiring an instruction set to be tested, wherein the instruction set to be tested comprises a plurality of instruction subsets;
step3: selecting any unselected instruction subset from the instruction set to be tested, sending the instruction subset to instrument equipment, carrying out image acquisition on the instrument equipment to obtain a feedback image set, and simultaneously interpreting the instruction subset through the terminal server to obtain detection contents;
step4: identifying a feedback atlas by using the panel image as a reference through a detection model to obtain difference characteristics, and matching the difference characteristics with detection contents to obtain a judgment record;
step5: and repeating the steps in Step3 and Step4 until all the instruction subsets in the instruction set to be tested are sent to instrument equipment, acquiring a plurality of judgment records, generating a detection record according to the judgment records, and storing the detection record.
In a specific embodiment, the panel image may be further identified by a detection model to obtain the indicator light features, and the color and brightness detection may be performed on the indicator light features by separating RGB channels of the image, calculating the brightness, the mean, the variance, and the like of each channel of the image.
In another embodiment, the panel image may be preprocessed, and the preprocessed panel image is identified by the detection model to obtain the surface quality defect of the instrument device.
In a preferred embodiment, as in Step1, the main steps of constructing the detection model at the terminal server are as follows:
step11: constructing a convolutional neural network model, wherein the convolutional neural network model at least comprises an input layer, a convolutional layer, a pooling layer, a full-link layer and an output layer;
step12: acquiring training data, verification data and test data, wherein the training data at least comprises a symbol identification atlas, a segment type icon set, a needle type atlas and a text type atlas;
step13: inputting the training data into a convolutional neural network for training to obtain a detection model based on the convolutional neural network;
step14: and verifying the detection model by using the verification data, and testing the detection model by using the test data.
In a specific embodiment, a convolutional neural network model to be trained is constructed, the neural network model at least comprises an input layer, a multi-level convolutional layer, a pooling layer, a full connection layer and an activation layer, an instrument device image set is obtained through a large data platform, an image acquisition channel, a third-party platform and other channels, the instrument device image set is classified to obtain training data, verification data and test data which comprise a symbol identification image set, a segment icon set, a needle image set and a text image set, the training data is input into the convolutional neural network model to be trained, and the training data is preprocessed through the input layer, wherein the preprocessing mode comprises scaling, cutting, rotating, turning, normalization and the like, so that the consistency and reliability of the input data are guaranteed, and an image feature matrix is output after the input layer preprocesses the training data, so that the image processing result can be greatly optimized, and different data units are effectively prevented from being input; performing convolution operation on the convolution layer, extracting features of input data, calculating the inner product of the input data, and outputting a feature map, for example, a convolution kernel of 5 × 5 can analyze image features of 5 × 5 by one-time operation, the whole image is extracted by using the convolution kernel in a sliding window manner, and image feature analysis is performed on each pixel point of the whole image, so that the feature map is obtained; then, the whole matrix is reduced through the pooling layer, the number of nodes in the last full-connection layer is further reduced, and therefore the purposes of reducing parameters in the whole neural network and reducing calculation complexity and memory consumption are achieved; after the processing of the multi-turn convolutional layer and the pooling layer, the feature matrix set is vectorized through the full connection layer, the features extracted by the convolutional layer and the pooling layer are expanded, classification or regression tasks are performed through the full connection layer, finally, the vectorized features are converted into labels through the activation layer, a convolutional neural network-based model is obtained, and finally, the identification is performed through panel parameters on the network model instrument.
In another embodiment, a product atlas with appearance defects (such as surface defects like scratches, pits, stains and the like) can be collected through image acquisition or big data acquisition, and the convolutional neural network model is trained through the product atlas with appearance defects, so that the detection model can detect the surface defects of the instrument equipment.
In a preferred embodiment, as described in Step1, the main steps of constructing the instruction library at the terminal server are as follows:
step15: constructing an instruction library, constructing a plurality of instruction subsets in the instruction library, and creating a primary signal tag for each instruction subset, wherein detection target information is recorded in the primary signal tag;
step16: and constructing a plurality of test messages in the instruction subset according to the primary signal label, and creating a secondary signal label for each test message, wherein the secondary signal label records target state information, and the target recorded in the secondary signal label is the same as the target recorded in the primary signal label.
Further, in this embodiment, the test packet is constructed in the instruction subset by network collection, self-compiling, and the like;
in one embodiment, the instruction library may be set up with reference to the following form:
specifically, the construction form of the instruction library may be adjusted according to the actual use requirement and the use scenario, and is not specifically limited herein.
In a preferred embodiment, as in Step2, the main steps of obtaining the instruction set to be tested are as follows:
step21: acquiring an image of instrument equipment through an image acquisition module to obtain a panel image;
step22: identifying the panel image through the detection model, acquiring panel parameters containing a plurality of panel characteristics, and creating an image label for each panel characteristic, wherein panel characteristic information is recorded in the image label;
step23: and calling the instruction subset matched with the image tag and the primary signal tag according to the image tag matching instruction library to obtain an instruction set to be tested.
In a specific embodiment, a panel image is input into a detection model, the panel image is preprocessed, convolution operation is carried out through convolution layers, features of input data are extracted, inner products of the features are calculated, a plurality of feature maps are output, and the expression capability of the features is enhanced by using an activation function, wherein each feature map corresponds to a filter and represents a certain feature detected by the filter on the input image, and the feature size of the image is reduced by carrying out maximum pooling or average pooling operation through a pooling layer; and repeating the convolutional layer, the activation function and the pooling layer to extract features of higher levels, after convolution, activation and pooling for multiple times, the network expands the feature map of the last layer into a one-dimensional vector and inputs the one-dimensional vector into the fully-connected layer so as to map the high-dimensional features to the probability distribution of the prediction label, and finally, the output layer maps the output of the fully-connected layer into a probability distribution through a Softmax function to represent the prediction probability of each category.
In a preferred embodiment, as in Step3, the main steps of acquiring an image of the meter device and obtaining a feedback atlas are as follows:
step31: selecting any unselected instruction subset from the instruction set to be tested, and sending a plurality of test messages in the instruction subset to instrument equipment one by one;
step32: and when one test message is sent, the image acquisition module acquires an image of the instrument equipment once to acquire a feedback image, and acquires a feedback image set until all the test messages in the instruction subset are sent.
In a specific embodiment, an instruction subset with a first-level signal label as a high beam is selected from the instruction set to be tested, the instruction subset comprises two test messages, the test message with a second-level signal label as the high beam is sent to instrument equipment, a first feedback image is obtained through an image acquisition module, the test message with the second-level signal label as the high beam is sent to the instrument equipment, a second feedback image is obtained through the image acquisition module, and the first feedback image and the second feedback image are stored in one atlas to obtain the feedback atlas.
In a preferred embodiment, as in Step4, the main steps of acquiring the determination record are as follows:
step41: inputting the feedback image and the panel image into a detection model, and acquiring the difference characteristics of the feedback image and the panel image;
step42: matching the difference characteristics with the instruction content to obtain a judgment record, wherein the judgment record comprises a feedback image, the instruction content and a judgment result;
step43: if the difference characteristics are not matched with the instruction content, judging that the result is unqualified;
step44: and if the difference characteristics are matched with the instruction content, judging that the result is qualified.
In a specific embodiment, after a feedback diagram set of the high beam is obtained by sending a command subset, the feedback diagram set of the high beam and a panel image are input to a monitoring model, each sub-element in the feedback diagram set is compared with the panel image respectively, RGB channel separation is performed on a first feedback image, the difference characteristic of the first feedback image and the panel image is obtained to indicate that color difference exists in the high beam identification, a test message with a secondary signal label for opening the high beam is interpreted, the difference characteristic is matched with the content of the message, the judgment result is qualified, and the feedback image, the test message and the judgment result are stored.
In a preferred embodiment, as stated in S5, the step after generating a detection record according to a plurality of determination records and storing the detection record as stated in S5 further includes:
s51: counting the qualified rate of the instrument equipment in unit time;
s52: acquiring the qualification rate in all detection periods, and inputting the qualification rate into a trend change function to obtain a change trend value of the qualification rate of the instrument equipment; the trend change function of the qualified rate is as follows:
in, is greater than or equal to>
Represents a change trend value, and>
representing the number of detection periods>
Represents the interval 2 to>
The qualification rate in (4), based on the status of the status quo>
Represents the interval 1 to>
-a yield in 1;
s53: obtaining an allowable deviation interval and comparing the allowable deviation interval with the change trend value;
s54: if the variation trend value is within the allowable deviation interval, judging that the production line of the instrument equipment runs normally;
s55: and if the variation trend value is lower than the lower limit value of the allowable deviation interval, judging that the production line of the instrument equipment is abnormal in operation, and sending an alarm signal.
In the embodiment, when the instrument equipment is detected, a multi-cycle qualification rate comparison method is adopted, the running state of an instrument equipment production line is indirectly reflected, so that problems can be found and solved in time, in the process, the qualification rates in a plurality of cycles are counted, and a trend change function is used for obtaining a change trend value of the qualification rate, because the production equipment can generate self-loss or aging after running for a long time, the corresponding output qualification rate is inevitably generated when the yield qualification rate is in a descending trend, and at the moment, the alarm signal is obviously not required to be sent, on the basis, the embodiment can deal with the phenomenon by presetting an allowable deviation interval, namely when the change trend value is lower than the lower limit value of the allowable deviation interval, the instrument equipment production line is judged to be abnormal, and correspondingly sent out the alarm signal to remind a maintenance worker to carry out maintenance operation, otherwise, the change trend value is in the allowable deviation interval, and the default is caused by the self-loss, aging and other reasons of the equipment, so that false alarm phenomenon is effectively avoided;
in addition, it should be noted that, since the loss of the device itself will cause the variation trend value (in the case of long-term operation of the same device, the yield is inevitably smaller and smaller, and the value of the variation trend value is negative), and there will also be a phenomenon below the allowable deviation interval, but the period is relatively long, and at this time, an alarm signal will still be sent, so when determining the yield input to the trend change function, the number of periods needs to be explicitly detected, and the occurrence of the phenomenon that the variation trend value is below the allowable deviation interval due to too much yield substituted into the trend change function is avoided.
In a preferred real-time manner, the step of obtaining the value of the variation trend of the qualified rate of the meter device further includes:
s56: acquiring a yield change value in adjacent detection periods in real time;
s57: acquiring a fluctuation threshold value, and comparing the fluctuation threshold value with the qualified rate change value;
s58: if the qualified rate change value is higher than or equal to the fluctuation threshold value, judging that the production line of the instrument equipment is abnormal and not adding the abnormal production line into the trend change function;
s59: and if the qualified rate change value is lower than a fluctuation threshold value, adding the qualified rate change value into a trend change function.
In the embodiment, within the allowable deviation interval, the phenomenon of instantaneous fluctuation still exists, the phenomenon of batch unqualified instrument production can be caused due to the fact that the probability of the instantaneous fluctuation corresponds to equipment faults under the condition that the equipment faults do not reach the alarm standard, and on the basis of the phenomenon, adjacent qualified rates are compared in advance through a preset fluctuation threshold value, whether the qualified rate change value exceeds the preset fluctuation threshold value or not is judged, if the qualified rate change values in two adjacent detection periods are smaller than or equal to the fluctuation threshold value, the qualified rate change is caused by normal loss or aging of a production line and belongs to a normal loss range, the change value of the qualified rate is normally added into a trend change function, and the change trend value of the qualified rate is continuously evaluated; if the qualification rate change values in two adjacent detection periods are larger than the fluctuation threshold value, the production line is abnormal in operation, an emergency occurs, and the qualification rate change value is not added into the trend change function in order to avoid influencing the judgment of the qualification rate change trend.
The instrument detection device based on visual recognition is applied to the instrument detection method based on visual recognition, and comprises the following steps:
the terminal server is used for constructing a detection model and an instruction library;
the image acquisition module is used for acquiring images of the instrument equipment;
the judging module is used for generating a judging record according to the instruction content and the feedback atlas;
the alarm module is used for sending out an alarm signal;
an interpretation module capable of interpreting the test message.
Referring to fig. 2, a server device, which may be a computer, a server or other terminal having data processing capability, is shown. The server device comprises a processor, a storage element, a communication module and an interpretation module connected by a system bus, the storage element storing a computer program capable of being run by the processor. The processor at least comprises a CPU, a memory, a BIOS chip and an I/O control chip, wherein the CPU is used for processing instructions, executing operations, requiring actions, controlling time and processing data, the memory element is used for temporarily storing operation data in the CPU and data exchanged with external storage elements such as a hard disk and the like, the BIOS chip is suitable for initialization and detection of various hardware devices in the starting process of the computer, the I/O control chip is used for managing all input and output devices of the system, and the interpretation module can interpret test messages and then test the detection content of the test messages. The storage element of the server device includes a non-volatile storage medium, an internal storage element. The non-volatile storage medium stores an operating system and a plurality of instructions. The memory element provides an environment for the operation of the operating system and instructions in the non-volatile storage medium. The computer program is executed by a processor to implement the steps of any one of the visual recognition-based meter detection methods.
A non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of a method of visual recognition based meter detection as described above.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above may be implemented by hardware instructions of a computer program, which may be stored in a non-volatile computer-readable storage medium, and when executed, may include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium provided herein and used in the examples may include non-volatile and/or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRDRAM), enhanced SDRAM (ESDRAM), synchronous Link (Synchlink) DRAM (SLDRAM), rambus (Rambus) direct RAM (RDRAM), direct bused dynamic RAM (DRDRAM), and bused dynamic RAM (RDRAM)
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and amendments can be made without departing from the principle of the present invention, and these modifications and amendments should also be considered as the protection scope of the present invention. Structures, devices, and methods of operation not specifically described or illustrated herein are generally practiced in the art without specific recitation or limitation.
Claims (10)
1. A meter detection method based on visual identification is characterized in that: the method mainly comprises the following steps:
s1: a detection model and an instruction library are established in a terminal server, wherein the instruction library comprises a plurality of instruction subsets;
s2: acquiring an image of instrument equipment through an image acquisition module to obtain a panel image, identifying the panel image through the detection model to obtain panel parameters, and matching an instruction base according to the panel parameters to obtain an instruction set to be tested;
s3: selecting any unselected instruction subset from the instruction set to be tested, sending the instruction subset to instrument equipment, carrying out image acquisition on the instrument equipment to obtain a feedback image set, and simultaneously interpreting the instruction subset through the terminal server to obtain detection contents;
s4: identifying a feedback atlas by using the panel image as a reference through a detection model to obtain difference characteristics, and matching the difference characteristics with detection contents to obtain a judgment record;
s5: and repeating the steps in the S3 and the S4 until all the instruction subsets in the instruction set to be tested are sent to the instrument equipment, acquiring a plurality of judgment records, generating a detection record according to the judgment records, and storing the detection record.
2. The meter detection method based on visual recognition according to claim 1, wherein: as in S1, the main steps of constructing the detection model at the terminal server are as follows:
s11: constructing a convolutional neural network model, wherein the convolutional neural network model at least comprises an input layer, a convolutional layer, a pooling layer, a full-link layer and an output layer;
s12: acquiring training data, verification data and test data, wherein the training data at least comprises a symbol identification atlas, a segment type icon atlas, a needle type atlas and a text type atlas;
s13: inputting the training data into a convolutional neural network for training to obtain a detection model based on the convolutional neural network;
s14: and verifying the detection model by using the verification data, and testing the detection model by using the test data.
3. The meter detection method based on visual recognition according to claim 1, wherein: as described in S1, the main steps of constructing the instruction library in the terminal server are as follows:
s15: constructing an instruction library, constructing a plurality of instruction subsets in the instruction library, and creating a primary signal label for each instruction subset, wherein detection target information is recorded in the primary signal label;
s16: and constructing a plurality of test messages in the instruction subset according to the primary signal tags, and creating a secondary signal tag for each test message, wherein the secondary signal tag records target state information.
4. The meter detection method based on visual recognition according to claim 1, wherein: as in S2, the main steps of obtaining the instruction set to be tested are as follows:
s21: acquiring an image of instrument equipment through an image acquisition module to obtain a panel image;
s22: identifying the panel image through the detection model, acquiring panel parameters containing a plurality of panel characteristics, and creating an image label for each panel characteristic, wherein panel characteristic information is recorded in the image label;
s23: and calling the instruction subset matched with the image tag and the primary signal tag according to the image tag matching instruction library to obtain an instruction set to be tested.
5. The meter detection method based on visual recognition according to claim 1, wherein: as in S3, the main steps of acquiring an image of the meter device and obtaining a feedback atlas are as follows:
s31: selecting any unselected instruction subset from the instruction set to be tested, and sending a plurality of test messages in the instruction subset to the instrument equipment one by one;
s32: and when one test message is sent, the image acquisition module acquires an image of the instrument equipment once to acquire a feedback image, and acquires a feedback image set until all test messages in the instruction subset are sent.
6. The meter detection method based on visual recognition according to claim 5, wherein: as in S4, the main steps of acquiring the determination record are as follows:
s41: inputting the feedback image and the panel image into a detection model, and acquiring the difference characteristics of the feedback image and the panel image;
s42: matching the difference characteristics with the instruction content to obtain a judgment record, wherein the judgment record comprises a feedback image, a test message, a secondary signal label and a judgment result;
s43: if the difference characteristics are not matched with the message content, judging that the result is unqualified;
s44: and if the difference characteristics are matched with the message content, judging that the result is qualified.
7. The meter detection method based on visual recognition according to claim 1, wherein: as described in S5, the step after generating a detection record from a plurality of the determination records and storing the detection record further includes:
s51: counting the qualification rate of the instrument equipment in unit time;
s52: acquiring the qualification rate in all detection periods, and inputting the qualification rate into a trend change function to obtain a change trend value of the qualification rate of the instrument equipment; the trend change function of the qualified rate is as follows:
in, is greater than or equal to>
Represents a change trend value>
Representing the number of detection periods>
Represents the interval 2 to>
The qualification rate in (4), based on the status of the status quo>
Represents the interval 1 to>
-a yield in 1;
s53: obtaining an allowable deviation interval and comparing the allowable deviation interval with the change trend value;
s54: if the variation trend value is within the allowable deviation interval, judging that the production line of the instrument equipment normally operates;
s55: and if the variation trend value is lower than the lower limit value of the allowable deviation interval, judging that the production line of the instrument equipment is abnormal in operation, and sending an alarm signal.
8. The meter inspection method based on visual recognition according to claim 7, wherein: the step after obtaining the variation trend value of the qualified rate of the instrument equipment further comprises the following steps:
s56: acquiring a qualification rate change value in the adjacent detection period in real time;
s57: acquiring a fluctuation threshold value, and comparing the fluctuation threshold value with the qualification rate change value;
s58: if the qualified rate change value is higher than or equal to the fluctuation threshold value, judging that the production line of the instrument equipment is abnormal and not adding the abnormal production line into the trend change function;
s59: and if the qualified rate change value is lower than a fluctuation threshold value, adding the qualified rate change value into a trend change function.
9. A meter inspection device based on visual recognition, which is applied to the meter inspection method based on visual recognition of any one of claims 1 to 8, wherein: the method comprises the following steps:
the terminal server is used for constructing a detection model and an instruction library;
the image acquisition module is used for acquiring images of the instrument equipment;
the judging module is used for generating a judging record according to the instruction content and the feedback atlas;
the alarm module is used for sending out an alarm signal;
an interpretation module capable of interpreting the test message.
10. A server device, characterized by: the server device comprising a processor, a memory element and a computer program stored on the memory element and executable by the processor, wherein the processor when executing the computer program performs the steps of a method for meter inspection based on visual recognition according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310252271.5A CN115965625B (en) | 2023-03-16 | 2023-03-16 | Instrument detection device based on visual recognition and detection method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310252271.5A CN115965625B (en) | 2023-03-16 | 2023-03-16 | Instrument detection device based on visual recognition and detection method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115965625A true CN115965625A (en) | 2023-04-14 |
| CN115965625B CN115965625B (en) | 2023-05-30 |
Family
ID=85888212
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310252271.5A Active CN115965625B (en) | 2023-03-16 | 2023-03-16 | Instrument detection device based on visual recognition and detection method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115965625B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116481585A (en) * | 2023-06-26 | 2023-07-25 | 广州晨安网络科技有限公司 | Product full detection system and method based on machine vision |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102628854A (en) * | 2012-03-29 | 2012-08-08 | 赵宝明 | Automobile instrument board detection system and method |
| CN113239928A (en) * | 2021-05-11 | 2021-08-10 | 北京百度网讯科技有限公司 | Method, apparatus and program product for image difference detection and model training |
| CN113865636A (en) * | 2021-09-28 | 2021-12-31 | 北斗星通智联科技有限责任公司 | Instrument testing method and system and electronic equipment |
| WO2023001169A1 (en) * | 2021-07-22 | 2023-01-26 | 中国第一汽车股份有限公司 | Automobile instrument test method, apparatus and system, and storage medium |
-
2023
- 2023-03-16 CN CN202310252271.5A patent/CN115965625B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102628854A (en) * | 2012-03-29 | 2012-08-08 | 赵宝明 | Automobile instrument board detection system and method |
| CN113239928A (en) * | 2021-05-11 | 2021-08-10 | 北京百度网讯科技有限公司 | Method, apparatus and program product for image difference detection and model training |
| WO2023001169A1 (en) * | 2021-07-22 | 2023-01-26 | 中国第一汽车股份有限公司 | Automobile instrument test method, apparatus and system, and storage medium |
| CN113865636A (en) * | 2021-09-28 | 2021-12-31 | 北斗星通智联科技有限责任公司 | Instrument testing method and system and electronic equipment |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116481585A (en) * | 2023-06-26 | 2023-07-25 | 广州晨安网络科技有限公司 | Product full detection system and method based on machine vision |
| CN116481585B (en) * | 2023-06-26 | 2023-09-12 | 广州晨安网络科技有限公司 | Product full detection system and method based on machine vision |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115965625B (en) | 2023-05-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112115927B (en) | Intelligent machine room equipment identification method and system based on deep learning | |
| CN109739904B (en) | Time sequence marking method, device, equipment and storage medium | |
| JP6871877B2 (en) | Information processing equipment, information processing methods and computer programs | |
| CN112487356B (en) | Structural health monitoring data enhancement method | |
| CN115965625B (en) | Instrument detection device based on visual recognition and detection method thereof | |
| CN116563762A (en) | Fire detection method, system, medium, equipment and terminal for oil and gas station | |
| CN112529109A (en) | Unsupervised multi-model-based anomaly detection method and system | |
| CN117351271A (en) | Fault monitoring method and system for high-voltage distribution line monitoring equipment and storage medium thereof | |
| CN115171045A (en) | YOLO-based power grid operation field violation identification method and terminal | |
| CN113762330A (en) | Method and device for optimizing service configuration in dynamic network | |
| CN117689646A (en) | High-precision defect detection method, system and medium for positive and negative sample fusion | |
| CN110817674B (en) | Method, device and equipment for detecting step defect of escalator and storage medium | |
| CN116843677A (en) | Appearance quality detection system and method for sheet metal part | |
| CN116188445A (en) | Product surface defect detection and positioning method and device and terminal equipment | |
| CN113592864B (en) | Transformer monitoring method, device, system and medium based on convolutional neural network | |
| CN115620083A (en) | Model training method, face image quality evaluation method, device and medium | |
| CN112149579A (en) | Improved yolo _ v 3-based bird nest hidden danger detection method | |
| CN110728310A (en) | Target detection model fusion method and system based on hyper-parameter optimization | |
| CN116863481B (en) | Service session risk processing method based on deep learning | |
| CN117984024B (en) | Welding data management method and system based on automatic production of ship lock lambdoidal doors | |
| KR102717943B1 (en) | Apparatus and method for detecting ai abnormalities in a plastic injection process | |
| CN112307090B (en) | Detection method and system based on tunnel security data | |
| CN118091406B (en) | Motor detection and repair method and device, electronic equipment and storage medium | |
| CN116740472B (en) | Power switch cabinet switch state identification method, device, equipment and storage medium | |
| CN117853774A (en) | Energy storage system abnormality identification method and system |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |