Method and device for normative inspection based on artificial intelligence

Abstract

The embodiment of the invention discloses a normativity inspection method and device based on artificial intelligence, and the method comprises the steps: firstly receiving and responding to a target detection instruction, selecting one or more target detection objects based on the target detection instruction, and better detecting operation objects or operation actions corresponding to the target detection objects. Target detection areas corresponding to the target detection objects are obtained in advance, and the target detection areas and the target detection objects are bound to form a standard inspection target; image data, including static picture data and dynamic video tracking data, are acquired through an image acquisition device, and the image data are preprocessed to obtain data with higher accuracy, so that feature points can be better extracted, and a target can be accurately identified; the embodiment of the invention greatly improves the monitoring feasibility and flexibility of the student experiment process on the basis of artificial intelligence, helps to accurately judge the operability of the student experiment, and improves the efficiency.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based standardization inspection method and device. Background Art

[0002] At present, the country is vigorously developing junior high school physics, chemistry and biology experiments to improve students' hands-on practical ability. Therefore, experiments are not only a verification tool for scientific theories, but also a necessary link for students to absorb knowledge and improve their practical ability. However, the laboratory environment is complex and changeable, and there are many types of instruments and equipment involved, which places strong requirements on the hands-on ability of students. Traditional experimental teaching and operation management models often have the following significant problems: First, due to the complexity of the instruments and cumbersome operation procedures, especially for beginners, it is easy to cause misoperation. For example, if the assembly of experimental equipment does not meet the specifications, it may lead to experimental failure or inaccurate results, and even cause safety accidents in some cases. Secondly, for teachers, facing a large number of students' experimental operations, the timeliness and accuracy of their teaching are limited. Teachers cannot effectively supervise and guide each experimental link, resulting in low quality of students' experimental learning. Summary of the invention

[0003] In response to the above-mentioned defects, an embodiment of the present invention discloses a standardization inspection method and device based on artificial intelligence, which can help improve the safety and standardization of experimental operations through intelligent and automatic detection.

[0004] The first aspect of the embodiment of the present invention discloses a standardization inspection method based on artificial intelligence, including:

[0005] In response to the target detection instruction, a target detection object is selected, a target detection area corresponding to the target detection object is acquired, and the target detection area and the target detection object are associated to obtain a standard inspection target;

[0006] Collect image data of the standard inspection target, the image data including picture data and video tracking data, preprocess the image data, extract feature points from the preprocessed image data, and obtain recognition results of the standard inspection target based on the feature points, the recognition results including detection standards, positions and types.

[0007] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the preprocessing of the image data includes:

[0008] Denoising the image data based on bilateral filtering;

[0009] Perform contrast adjustment on the denoised image data;

[0010] The size of the image data is adjusted, and an interference area in the image data is identified and the interference area is cropped.

[0011] As an optional implementation manner, in the first aspect of the embodiment of the present invention, extracting feature points from the preprocessed image data includes:

[0012] The preprocessed image data is input into the preset YOLO deep learning model to extract the feature points of the image data.

[0013] As an optional implementation manner, in the first aspect of the embodiment of the present invention, extracting feature points from the preprocessed image data further includes:

[0014] The preprocessed image data is input into the preset hand key point detection model and eye key point detection model to extract feature points.

[0015] As an optional implementation manner, in the first aspect of the embodiment of the present invention, obtaining the recognition result of the standard inspection target according to the feature point includes:

[0016] Divide the image data into multiple identification areas, and obtain the code corresponding to each identification area according to the feature points;

[0017] Acquire the wire connection information between each two identification areas to create a directed graph adjacency table, and acquire the preset current flow direction of each identification area, and add a node in the directed graph adjacency table based on the preset current flow direction;

[0018] Based on the directed graph adjacency table after adding nodes, it is determined whether there is a current loop in the image data to obtain circuit connection information of the image data, and a corresponding recognition result is generated according to the circuit connection information.

[0019] As an optional implementation manner, in the first aspect of the embodiment of the present invention, it also includes:

[0020] When the movement of the standard inspection target is detected based on the visual tracking data, the moving image of the standard inspection target is extracted at preset intervals to generate a change path, and whether to generate warning information is determined according to the change path.

[0021] As an optional implementation, in the first aspect of the embodiment of the present invention, when the movement of the standard inspection target is detected based on the visual tracking data, the moving image of the standard inspection target is extracted at preset intervals to generate a change path, including:

[0022] When the movement of the specification inspection target is detected based on the visual tracking data, the video tracking data is cut according to a preset interval to be divided into several segments of moving images;

[0023] Identify and mark the standard inspection target in each moving image, and obtain the standard inspection target position information in each video clip;

[0024] A change path of the standard inspection target is generated based on the standard inspection target position information of two adjacent moving images.

[0025] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the determining whether to generate warning information according to the change path includes:

[0026] Acquire the change type of the specification inspection target according to the change path, and select the corresponding specification inspection target transformation benchmark according to the change type;

[0027] The change path is compared with the standard inspection target transformation benchmark to calculate the matching degree between the change path and the standard inspection target transformation benchmark, and when the matching degree is lower than a preset value, an early warning message is generated.

[0028] A second aspect of an embodiment of the present invention discloses a standardization inspection device based on artificial intelligence, comprising:

[0029] Instruction response module: used to select a target detection object in response to a target detection instruction, obtain a target detection area corresponding to the target detection object, and associate the target detection area with the target detection object to obtain a standard inspection target;

[0030] Image acquisition module: used to collect image data of the standard inspection target, the image data includes picture data and video tracking data, preprocess the image data, extract feature points from the preprocessed image data, and obtain the recognition result of the standard inspection target based on the feature points, the recognition result includes the detection standard, position and type.

[0031] As an optional implementation manner, in the second aspect of the embodiment of the present invention, the preprocessing of the image data includes:

[0032] Denoising the image data based on bilateral filtering;

[0033] Perform contrast adjustment on the denoised image data;

[0034] The size of the image data is adjusted, and an interference area in the image data is identified and the interference area is cropped.

[0035] As an optional implementation manner, in the second aspect of the embodiment of the present invention, extracting feature points from the preprocessed image data includes:

[0036] The preprocessed image data is input into the preset YOLO deep learning model to extract the feature points of the image data.

[0037] As an optional implementation manner, in the second aspect of the embodiment of the present invention, extracting feature points from the preprocessed image data further includes:

[0038] The preprocessed image data is input into the preset hand key point detection model and eye key point detection model to extract feature points.

[0039] As an optional implementation manner, in the second aspect of the embodiment of the present invention, obtaining the recognition result of the standard inspection target according to the feature point includes:

[0040] Divide the image data into multiple identification areas, and obtain the code corresponding to each identification area according to the feature points;

[0041] Acquire the wire connection information between each two identification areas to create a directed graph adjacency table, and acquire the preset current flow direction of each identification area, and add a node in the directed graph adjacency table based on the preset current flow direction;

[0042] Based on the directed graph adjacency table after adding nodes, it is determined whether there is a current loop in the image data to obtain circuit connection information of the image data, and a corresponding recognition result is generated according to the circuit connection information.

[0043] As an optional implementation, the second aspect of the embodiment of the present invention also includes a target comparison module: when the movement of the standard inspection target is detected based on visual tracking data, the moving image of the standard inspection target is extracted at preset intervals to generate a change path, and whether to generate warning information based on the change path.

[0044] As an optional implementation, in the second aspect of the embodiment of the present invention, when the movement of the standard inspection target is detected based on the visual tracking data, the moving image of the standard inspection target is extracted at preset intervals to generate a change path, including:

[0045] When the movement of the specification inspection target is detected based on the visual tracking data, the video tracking data is cut according to a preset interval to be divided into several segments of moving images;

[0046] Identify and mark the standard inspection target in each moving image, and obtain the standard inspection target position information in each video clip;

[0047] A change path of the standard inspection target is generated based on the standard inspection target position information of two adjacent moving images.

[0048] As an optional implementation manner, in the second aspect of the embodiment of the present invention, the determining whether to generate warning information according to the change path includes:

[0049] Acquire the change type of the specification inspection target according to the change path, and select the corresponding specification inspection target transformation benchmark according to the change type;

[0050] The change path is compared with the standard inspection target transformation benchmark to calculate the matching degree between the change path and the standard inspection target transformation benchmark, and when the matching degree is lower than a preset value, an early warning message is generated.

[0051] The third aspect of an embodiment of the present invention discloses an electronic device, comprising: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the artificial intelligence-based standardization inspection method disclosed in the first aspect of the embodiment of the present invention.

[0052] A fourth aspect of an embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the artificial intelligence-based standardization checking method disclosed in the first aspect of an embodiment of the present invention.

[0053] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0054] In the embodiment of the present invention, a target detection instruction is first received and responded to, and one or more target detection objects are selected based on the target detection instruction. In order to better detect the operating items or operating actions corresponding to the target detection objects, the target detection areas corresponding to these target detection objects are obtained in advance, and the two are bound to form a standardized inspection target; image data is collected by an image acquisition device, including static picture data and dynamic video tracking data. After preprocessing, the image data is obtained to obtain data with higher accuracy, which is helpful to better extract feature points and then accurately identify the target; the embodiment greatly improves the feasibility and flexibility of monitoring the student experimental process based on artificial intelligence, helps to accurately judge the operability of the student experiment, greatly saves manpower and improves efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0056] Figure 1It is a flowchart of a standardization inspection method based on artificial intelligence disclosed in an embodiment of the present invention;

[0057] Figure 2 It is a structural schematic diagram of a standardization inspection device based on artificial intelligence provided by an embodiment of the present invention;

[0058] Figure 3 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] It should be noted that the terms "first", "second", "third", "fourth", etc. in the specification and claims of the present invention are used to distinguish different objects rather than to describe a specific order. The terms "including" and "having" in the embodiments of the present invention and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0061] The embodiments of the present invention disclose a standard inspection method, device, electronic device and storage medium based on artificial intelligence. In the embodiments, a target detection instruction is first received and responded to, and one or more target detection objects are selected based on the target detection instruction. In order to better detect the operating items or operating actions corresponding to the target detection objects, target detection areas corresponding to these target detection objects are obtained in advance, and the two are bound to form a standard inspection target; image data is collected by an image acquisition device, including static picture data and dynamic video tracking data. After preprocessing, the image data obtains data with higher accuracy, which is helpful to better extract feature points, and then accurately identify the target, generate a dynamic change path of the target, and determine whether an early warning is needed according to the change path; the embodiments greatly improve the feasibility and flexibility of monitoring the student experimental process on the basis of artificial intelligence, help accurately judge the operability of the student experiment, greatly save manpower and improve efficiency.

[0062] Embodiment 1

[0063] See also Figure 1 , Figure 1It is a flow chart of the AI-based normative inspection method disclosed in the embodiment of the present invention. Among them, the execution subject of the method described in the embodiment of the present invention is an execution subject composed of software and / or hardware, which can receive relevant information by wired or / and wireless means, and can send certain instructions. Of course, it can also have certain processing functions and storage functions. The execution subject can control multiple devices, such as a remote physical server or cloud server and related software, or it can be a local host or server and related software that performs related operations on a device placed somewhere. In some scenarios, multiple storage devices can also be controlled, and the storage devices can be placed in the same place or different places as the devices. For example Figure 1 As shown, the AI-based normative inspection method includes the following steps:

[0064] 101. In response to a target detection instruction, a target detection object is selected, a target detection area corresponding to the target detection object is acquired, and the target detection area and the target detection object are associated to obtain a standard inspection target.

[0065] In the embodiment, the target detection instruction may be user input, a preset rule or other external system, etc. According to the instruction, one or more target detection objects may be selected. Usually, the target detection object is a person. When applied in a student laboratory scenario, the target detection object is usually a student. At the same time, a target detection area related to the target detection object is also obtained. The target detection area is closely related to the target detection object. For example, the target detection area may be a body part in the target detection object, or it may be an instrument operated by the target detection object. The target detection area may be one or more. Exemplarily, when the target detection area is an experimental instrument, there may be multiple experimental instruments of the same type and they are placed in different positions, but one of them is for the target detection object to operate, so the target detection area is bound to the target detection object.

[0066] 102. Collect image data of the standard inspection target, the image data including picture data and video tracking data, preprocess the image data, extract feature points from the preprocessed image data, and obtain recognition results of the standard inspection target based on the feature points, the recognition results including detection standards, positions and types.

[0067] In the embodiments, computer vision technology is used for image acquisition and preprocessing to ensure that the image quality meets the requirements of subsequent analysis; deep learning models are trained and optimized, and the performance of the models directly affects the recognition accuracy of clothing features; the judgment of detection results is based on preset safety clothing rules to accurately evaluate the compliance of clothing; the formulation and implementation of safety clothing rules, formulate reasonable safety clothing rules, and ensure that they are effectively implemented.

[0068] The image data of the target for standard inspection is collected, and the data includes picture data and video tracking data, which provide rich information about the appearance and movement of the target object. The image data undergoes a preprocessing step to remove noise, enhance features, adjust contrast, etc., so as to improve the accuracy of subsequent feature extraction. Specifically, the preprocessing of the image data in this embodiment includes: denoising the image data based on bilateral filtering; adjusting the contrast of the denoised image data; adjusting the size of the image data, and identifying interference areas in the image data, and cropping the interference areas.

[0069] Bilateral filtering is a nonlinear, edge-preserving smoothing filtering method. It considers the weights of both the spatial domain (i.e., the physical distance between pixels) and the pixel value domain (i.e., the difference between pixel values). This method can effectively remove noise while retaining the edge information of the image. During operation, set the parameters of bilateral filtering, including the Gaussian standard deviation (σ_s) in the spatial domain and the Gaussian standard deviation (σ_r) in the pixel value domain, apply the bilateral filtering algorithm to the image data, and obtain the denoised image. Contrast adjustment is a common step in image processing, which aims to enhance the brightness difference between different regions in the image to make the image clearer and easier to distinguish. During operation, calculate the global or local contrast of the image, and adjust the brightness value of the image according to the contrast calculation result to increase or decrease the contrast. Contrast adjustment can be achieved by using methods such as histogram equalization and adaptive contrast enhancement. In image data, interference areas may include irrelevant backgrounds, occluders, noise, etc., which may affect the results of subsequent image analysis. Therefore, it is necessary to identify and crop these interference areas. Use methods such as image segmentation, edge detection, and color space analysis to identify interference areas. According to the recognition results, the location and range of the interference area are determined. The image is cropped to remove the interference area and retain the target area.

[0070] Furthermore, the denoising method in the embodiment can also achieve higher quality denoising by learning more complex noise patterns through a large amount of data training through convolutional networks or adversarial networks. A denoising network is constructed, and noisy images and clean images are used as training data to train the denoising network. After training, the denoising network is used to denoise the image data. For contrast enhancement and color correction processing, multiple preset areas in the image can be identified based on image segmentation technology, and different contrast enhancement and color correction strategies can be applied to each area.

[0071] Then, feature points are extracted from the preprocessed image data, including: inputting the preprocessed image data into a preset YOLO deep learning model to extract feature points of the image data.

[0072] It further includes: inputting the preprocessed image data into a preset hand key point detection model and an eye key point detection model to extract feature points.

[0073] As an application scenario, in the student laboratory, the main purpose is to capture the state of the experimental data machine and monitor whether the placement and use of a kind of equipment in the laboratory is safe. In the image acquisition, the installation position of the camera ensures that the entire area to be monitored is covered, and the image of each experimental equipment can be captured, so as to achieve comprehensive monitoring of the experimental process. In this application scenario, the preprocessing step is used to improve the accuracy and efficiency of subsequent target detection, including the use of filtering technology to remove noise in the image and ensure the clarity and quality of the image. At the same time, in order to meet the input size requirements of the model, the image data is scaled and resized to reduce the computational complexity and improve the processing speed. In the recognition link, the deep learning model YOLO is used to accurately detect the target of the experimental instruments, such as test tubes, beakers, microscopes, etc. The deep learning model can automatically identify different types of instruments and extract their categories and position coordinates. After the instrument is identified, the location information of each instrument and its key components will be obtained, which is the basis for subsequent normative judgment. According to the preset instrument assembly specifications, the method will make a comprehensive judgment on the identified experimental instruments. The judgment content includes whether the instrument is placed in the specified position, whether the various components of the instrument are firmly connected, and whether the instrument has deviations in the horizontal and vertical directions. These judgments not only help ensure the standardization of experimental operations, but also improve the safety of the experiment and avoid accidents caused by improper placement of instruments.

[0074] Furthermore, the method may further include step 103, when the movement of the standard inspection target is detected based on the visual tracking data, the moving image of the standard inspection target is extracted at preset intervals to generate a change path, and whether to generate warning information is determined according to the change path.

[0075] In the above application scenarios, the target tracking and recognition technology is introduced, that is, tracking the target through video and tracking algorithms, which can track the key points of instruments or operators in the laboratory in real time. This technology can accurately identify in the video stream, ensure that the status of the instrument is continuously monitored throughout the experiment, and timely discover and correct potential problems. ByteTrack technology can maintain accurate tracking of targets in complex experimental environments through efficient data association algorithms, effectively reducing recognition errors caused by factors such as lighting changes and occlusion.

[0076] In another scenario, the application and the laboratory jointly detect and judge the key points of the students' body parts. Specifically, the YOLO deep learning algorithm is also applicable for fast and accurate target detection. In addition to detecting experimental instruments, the resnet convolutional neural network is used to detect the key points of the hands. The key points can be set in advance at different positions of the face. For example, a total of 21 key points are extracted. It has high processing accuracy and fast speed, which is suitable for real-time applications. For the detection of facial key points, retinaface can be used to achieve 5-point facial key point detection, which has high processing accuracy and fast speed, and is suitable for real-time applications.

[0077] In the application scenario of body part detection, by calculating the coordinates of the center points of the narrow-necked bottle and the measuring cylinder, the geometric relationship is used to determine whether the mouth of the narrow-necked bottle is directly above the measuring cylinder. The key points of the hand (such as the fingertips) are detected within the boundary box of the narrow-necked bottle to ensure correct hand contact. By calculating the relative position of the key points of the eyes (such as the corners of the eyes) and the water level of the measuring cylinder, the distance is judged to ensure that students can visually see the water level.

[0078] In this step, the recognition result of the standard inspection target is obtained according to the feature points, including: dividing the image data into multiple recognition areas, and obtaining the code corresponding to each recognition area according to the feature points; obtaining the wire connection information between every two recognition areas to create a directed graph adjacency list, and obtaining the preset current flow direction of each recognition area, and adding nodes to the directed graph adjacency list based on the preset current flow direction; judging whether there is a current loop in the image data based on the directed graph adjacency list after adding the nodes, so as to obtain the circuit connection information of the image data, and generating the corresponding recognition result according to the circuit connection information.

[0079] The above steps are applied in another application scenario, that is, electrical experiments. Traditional electrical experiments mainly rely on manual judgment when checking circuit connections. There are many problems with this. For example, in experimental examinations, a large number of invigilators are required to check the circuit connections of each student one by one, which greatly increases the workload of manual scoring, and manual inspections may lead to misjudgments due to subjective factors or fatigue, affecting the fairness of the examination. At the same time, in daily teaching, when students perform experimental operations, teachers may find it difficult to fully take into account the operation of each student due to the large number of students, and cannot promptly discover the errors in students' circuit connections, which is not conducive to the effective development of teaching. In addition, traditional manual detection methods are inefficient and cannot quickly and accurately judge the circuit conditions of a large number of students. Moreover, manual detection can often only be carried out at a specific time and place, and real-time monitoring and feedback cannot be achieved. With the continuous development of science and technology, although some auxiliary detection methods have emerged, there are still many shortcomings. For example, some simple electronic detection equipment has a relatively single function and can only detect some basic circuit parameters. The detection equipment needs to be connected to the circuit where the students are doing experiments, which affects the normal experimental process of the students, and cannot fully and accurately judge the correctness of the circuit connection. Embodiments are applied in this scenario, and machine vision recognition technology can be used to identify equipment, and the connection status of the equipment terminals and wire ends can be analyzed in real time. First, a high-definition camera is used to obtain the connection images of equipment, terminals and wire ends in junior high school physics and electrical experiments. These cameras are carefully arranged in the experimental area to ensure that the images of each key part can be captured comprehensively and clearly. Secondly, the acquired images are processed using advanced image recognition algorithms. By extracting and analyzing the shape, color, position and other features of the equipment, terminals and wire ends in the image, their positions and states in the image are accurately identified. Then, the connection relationship of the identified equipment, terminals and wire ends is calculated and judged with the help of a powerful graph algorithm. This graph algorithm can accurately analyze whether the current circuit connection meets the experimental requirements based on the pre-set circuit connection rules and patterns. In this application scenario, taking the detection of a series circuit composed of a battery, a switch, a light bulb, and an ammeter as an example to further illustrate, a terminal device for collecting video is placed on the experimental table where the student is doing the experiment. The device has a camera, and the student's circuit connection status is collected in real time from just above the experimental table. The video stream is transmitted to the artificial intelligence analysis server through the network. The server is used to perform artificial intelligence analysis on the circuit connection status in the video stream to determine the circuit connection status. First, by identifying the equipment in the video, including but not limited to batteries, ammeters, switches, small light bulbs), the position, size and category of the equipment in the video are identified through the target detection algorithm. At the same time, the position and size of the positive and negative terminals of the equipment in the video are also detected.By identifying the position and size of the wire heads in the video, the wires used are ensured to be of different colors, such as (red, green, blue, black, etc.). Each color of wire head is a category, and each terminal is numbered to facilitate the subsequent algorithm description. The terminals of each device are modeled, and according to the intersection of the rectangular box of the wire head and the rectangular box of the terminal detected by artificial intelligence, it is determined that there is a wire connection between the two terminals, thereby creating a directed graph adjacency table. Then, according to the current flow direction allowed inside the equipment, nodes are added to the directed graph adjacency table. For example, the internal current flow direction of the ammeter can only flow from the positive terminal to the negative terminal. Switches and light bulbs do not limit the current flow direction. The current flow direction of the battery is from the negative terminal to the positive terminal. Starting from the positive terminal of the battery, a depth-first traversal is performed on the graph. If a loop is detected in the graph, it means that the circuit is connected into a loop. There is only one loop in the graph, which means that the circuit is a series circuit. The connection status of the series circuit can be known through the order of depth traversal to know the connection status of the equipment in the circuit. The connection status is replaced by the graph node name: battery positive terminal -> ammeter positive terminal -> ammeter negative terminal -> light bulb positive terminal -> light bulb negative terminal -> switch positive terminal -> switch negative terminal -> battery negative terminal.

[0080] In this step, when the movement of the standard inspection target is detected based on the visual tracking data, the moving image of the standard inspection target is extracted at preset intervals to generate a change path, including: when the movement of the standard inspection target is detected based on the visual tracking data, the video tracking data is cut at preset intervals to be divided into several segments of moving images; the standard inspection target in each segment of the moving image is identified and marked, and the position information of the standard inspection target in each segment of the video clip is obtained; the change path of the standard inspection target is generated based on the position information of the standard inspection target of two adjacent moving images. By generating a change path, for example, the opening action of the bottle cap, the transformation from opening to fully opening, and then the placement of the bottle cap, for example, the bottle cap should be placed in position A to be standard, but if the student places it in position B, there is a possibility of unsafety. Whether it meets the standards can be determined by the change path.

[0081] Furthermore, determining whether to generate a warning message is performed based on the change path, including: obtaining a change type of the specification inspection target based on the change path, and selecting a corresponding specification inspection target transformation benchmark based on the change type; comparing the change path with the specification inspection target transformation benchmark, and calculating a degree of match between the change path and the specification inspection target transformation benchmark, and generating a warning message when the degree of match is lower than a preset value.

[0082] Embodiment 2

[0083] See also Figure 2 , Figure 2Schematic diagram of a standard inspection device based on artificial intelligence disclosed in an embodiment of the present invention. Figure 2 As shown, the artificial intelligence-based normative inspection device may include: an instruction response module 201 and an image acquisition module 202, wherein the instruction response module 201 is used to select a target detection object in response to a target detection instruction, obtain a target detection area corresponding to the target detection object, and associate the target detection area with the target detection object to obtain a normative inspection target; the image acquisition module 202 is used to acquire image data of the normative inspection target, the image data includes picture data and video tracking data, pre-process the image data, extract feature points from the pre-processed image data, so as to obtain a recognition result of the normative inspection target according to the feature points, and the recognition result includes a detection standard, a position and a type.

[0084] Furthermore, it may also include a target comparison module 203: when the movement of the standard inspection target is detected based on the visual tracking data, the moving image of the standard inspection target is extracted at preset intervals to generate a change path, and whether to generate warning information according to the change path.

[0085] In the image acquisition module 202, the image data is preprocessed, including: denoising the image data based on bilateral filtering; adjusting the contrast of the denoised image data; adjusting the size of the image data, identifying the interference area in the image data, and cropping the interference area.

[0086] Furthermore, extracting feature points from the preprocessed image data includes: inputting the preprocessed image data into a preset YOLO deep learning model to extract feature points of the image data. Preferably, it also includes: inputting the preprocessed image data into a preset hand key point detection model and an eye key point detection model to extract feature points.

[0087] The method obtains the recognition result of the standard inspection target according to the feature points, including: dividing the image data into multiple recognition areas, and obtaining the code corresponding to each recognition area according to the feature points; obtaining the wire connection information between every two recognition areas to create a directed graph adjacency list, and obtaining the preset current flow direction of each recognition area, and adding nodes to the directed graph adjacency list based on the preset current flow direction; judging whether there is a current loop in the image data based on the directed graph adjacency list after adding the nodes, so as to obtain the circuit connection information of the image data, and generating the corresponding recognition result according to the circuit connection information.

[0088] In the target comparison module 203, when the movement of the standard inspection target is detected based on the visual tracking data, the moving image of the standard inspection target is extracted at preset intervals to generate a change path, including: when the movement of the standard inspection target is detected based on the visual tracking data, the video tracking data is cut at preset intervals to be divided into several segments of moving images; the standard inspection target in each segment of the moving image is identified and marked, and the position information of the standard inspection target in each segment of the video clip is obtained; and the change path of the standard inspection target is generated based on the position information of the standard inspection target of two adjacent moving images.

[0089] Further, judging whether to generate warning information is based on the change path, including: obtaining the change type of the standard inspection target according to the change path, and selecting the corresponding standard inspection target transformation benchmark according to the change type; comparing the change path with the standard inspection target transformation benchmark to calculate the matching degree between the change path and the standard inspection target transformation benchmark, and generating warning information when the matching degree is lower than a preset value.

[0090] Embodiment 3

[0091] See also Figure 3 , Figure 3 Schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention. The electronic device may be a computer, a server, etc. Of course, in certain circumstances, it may also be a smart device such as a mobile phone, a tablet computer, a monitoring terminal, and an image acquisition device with processing functions. Figure 3 As shown, the electronic device may include:

[0092] A memory 301 storing executable program codes;

[0093] a processor 302 coupled to the memory 301;

[0094] The processor 302 calls the executable program code stored in the memory 301 to execute part or all of the steps in the artificial intelligence-based standardization inspection method in Example 1.

[0095] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute part or all of the steps in the artificial intelligence-based standardization checking method in Embodiment 1.

[0096] An embodiment of the present invention further discloses a computer program product, wherein when the computer program product runs on a computer, the computer executes part or all of the steps in the artificial intelligence-based standardization checking method in embodiment one.

[0097] An embodiment of the present invention also discloses an application publishing platform, wherein the application publishing platform is used to publish a computer program product, wherein when the computer program product runs on a computer, the computer executes part or all of the steps in the artificial intelligence-based standardization inspection method in embodiment one.

[0098] In various embodiments of the present invention, it should be understood that the size of the serial numbers of the processes does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0099] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place or distributed over multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0100] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0101] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a memory and includes several requests for a computer device (which can be a personal computer, a server or a network device, etc., specifically a processor in a computer device) to perform some or all of the steps of the method described in each embodiment of the present invention.

[0102] In the embodiments provided by the present invention, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined according to A. However, it should also be understood that determining B according to A does not mean determining B only according to A, and B can also be determined according to A and / or other information.

[0103] A person of ordinary skill in the art can understand that some or all of the steps in the various methods of the embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0104] The above is a detailed introduction to the artificial intelligence-based normative inspection method, device, electronic device and storage medium disclosed in the embodiments of the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A standard inspection method based on artificial intelligence, characterized in that: include: In response to the target detection instruction, a target detection object is selected, a target detection area corresponding to the target detection object is acquired, and the target detection area and the target detection object are associated to obtain a standard inspection target; Collect image data of the standard inspection target, the image data including picture data and video tracking data, preprocess the image data, extract feature points from the preprocessed image data, and obtain recognition results of the standard inspection target based on the feature points, the recognition results including detection standards, positions and types.

2. The method for checking the compliance of claim 1, characterized in that: The preprocessing of the image data comprises: Denoising the image data based on bilateral filtering; Perform contrast adjustment on the denoised image data; The size of the image data is adjusted, and an interference area in the image data is identified and the interference area is cropped.

3. The method for checking the compliance of claim 1, wherein: Extract feature points from preprocessed image data, including: The preprocessed image data is input into the preset YOLO deep learning model to extract the feature points of the image data.

4. The method for checking the standardization according to claim 3, characterized in that: Extracting feature points from preprocessed image data also includes: The preprocessed image data is input into the preset hand key point detection model and eye key point detection model to extract feature points.

5. The method for checking the compliance of claim 1 or 2, characterized in that: The recognition result of the standard inspection target is obtained according to the feature points, including: Divide the image data into multiple identification areas, and obtain the code corresponding to each identification area according to the feature points; Acquire the wire connection information between each two identification areas to create a directed graph adjacency table, and acquire the preset current flow direction of each identification area, and add a node in the directed graph adjacency table based on the preset current flow direction; Based on the adjacency table of the directed graph after adding nodes, it is determined whether there is a current loop in the image data to obtain circuit connection information of the image data, and a corresponding recognition result is generated according to the circuit connection information.

6. The method for checking the compliance of claim 1, wherein: Also includes: When the movement of the standard inspection target is detected based on the visual tracking data, the moving image of the standard inspection target is extracted at preset intervals to generate a change path, and whether to generate warning information is determined according to the change path.

7. The method for checking the compliance of claim 6, wherein: When the movement of the standard inspection target is detected based on the visual tracking data, the moving image of the standard inspection target is extracted at preset intervals to generate a change path, including: When the movement of the specification inspection target is detected based on the visual tracking data, the video tracking data is cut according to a preset interval to be divided into several segments of moving images; Identify and mark the standard inspection target in each moving image, and obtain the standard inspection target position information in each video clip; A change path of the standard inspection target is generated based on the standard inspection target position information of two adjacent moving images.

8. The method for checking compliance according to claim 6, characterized in that: The determining whether to generate warning information according to the change path includes: Acquire the change type of the specification inspection target according to the change path, and select the corresponding specification inspection target transformation benchmark according to the change type; The change path is compared with the standard inspection target transformation benchmark to calculate the matching degree between the change path and the standard inspection target transformation benchmark, and when the matching degree is lower than a preset value, an early warning message is generated.

9. A standard inspection device based on artificial intelligence, characterized in that: include: Instruction response module: used to select a target detection object in response to a target detection instruction, obtain a target detection area corresponding to the target detection object, and associate the target detection area with the target detection object to obtain a standard inspection target; Image acquisition module: used to collect image data of the standard inspection target, the image data includes picture data and video tracking data, preprocess the image data, extract feature points from the preprocessed image data, and obtain the recognition result of the standard inspection target based on the feature points, the recognition result includes the detection standard, position and type.

10. An electronic device, characterized in that: include: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the artificial intelligence-based standardization checking method described in any one of claims 1 to 7.

Images