Injection molding part screw detection method and system based on deep learning model, and medium

Abstract

The invention discloses an injection molding part screw detection method and system based on a deep learning model and a medium, applied to an upper computer, a control module and a detection module, the method comprises the following steps: the upper computer inputs a received screw hole image of a to-be-detected injection molding part into a preset deep learning model to generate coordinate information of a to-be-detected screw hole, the information is sent to the control module; the control module moves the to-be-detected injection molding part to a target position according to the coordinate information of the to-be-detected screw hole so as to obtain an injection molding part in-place signal; and the control module controls the detection module to carry out magnetic detection on the injection molding part to be detected according to the injection molding part in-place signal so as to generate a corresponding injection molding part screw detection result. By implementing the method provided by the invention, the problem that the screw mounting conditions of all injection molded parts cannot be efficiently detected in the prior art can be solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent manufacturing, and particularly relates to a screw detection method and system for injection molding parts based on a deep learning model and a medium. BACKGROUND

[0002] With the rapid development of science and technology, air conditioner production technology is constantly innovated, and the level of automation and intelligence is increasingly improved. However, there are still significant problems in the detection of air conditioner injection molding part screws on the current market. In the air conditioner production process, screws are needed to fasten injection molding parts, but the screw installation position is often in the deep hole of the injection molding part, and the light is insufficient and difficult to observe, resulting in frequent screw missing and serious quality risks. The existing detection technology relies on manual detection of injection molding parts one by one to determine whether the screws are missing. This way not only is inefficient, but also consumes a lot of manpower. In addition, there are many types of injection molding parts, and the deep hole positions are not uniform, so traditional automation technology cannot achieve batch detection of all injection molding part screws, resulting in the existing technology being unable to efficiently detect the screw installation of all injection molding parts. SUMMARY

[0003] The embodiments of the present application provide a screw detection method and system for injection molding parts based on a deep learning model and a medium, aiming to solve the problem that the existing technology cannot efficiently detect the screw installation of all injection molding parts.

[0004] In a first aspect, the embodiments of the present application provide a screw detection method for injection molding parts based on a deep learning model, applied to a host computer, a control module and a detection module. The method comprises: inputting a screw hole image of a to-be-detected injection molding part received by the host computer into a preset deep learning model to generate to-be-detected screw hole coordinate information, and sending the to-be-detected screw hole coordinate information to the control module; moving the to-be-detected injection molding part to a target position according to the to-be-detected screw hole coordinate information to obtain an injection molding part in-place signal; and controlling the detection module to perform magnetic detection on the to-be-detected injection molding part according to the injection molding part in-place signal to generate a corresponding injection molding part screw detection result.

[0005] In a second aspect, the embodiments of the present application also provide a screw detection system for injection molding parts based on a deep learning model, comprising a host computer, a control module and a detection module. The host computer is configured to input a screw hole image of a to-be-detected injection molding part received by the host computer into a preset deep learning model to generate to-be-detected screw hole coordinate information, and send the to-be-detected screw hole coordinate information to the control module. The control module is configured to move the to-be-detected injection molding part to a target position according to the to-be-detected screw hole coordinate information to obtain an injection molding part in-place signal. The control module is configured to control the detection module to perform magnetic detection on the to-be-detected injection molding part according to the injection molding part in-place signal to generate a corresponding injection molding part screw detection result.

[0006] In a third aspect, an embodiment of the present application further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the above method when executing the computer program.

[0007] In a fourth aspect, an embodiment of the present application further provides a computer readable storage medium, wherein the storage medium stores a computer program, and the computer program comprises program instructions, and the program instructions can implement the above method when being executed by a processor.

[0008] The embodiment of the present application provides a screw detection method and system for injection molding parts based on a deep learning model, which are applied to an upper computer, a control module and a detection module. The method comprises the following steps: the upper computer inputs a screw hole image of an injection molding part to be detected received by the upper computer into a preset deep learning model to generate coordinate information of a screw hole to be detected, and sends the coordinate information to the control module; the control module moves the injection molding part to be detected to a target position according to the coordinate information of the screw hole to be detected to obtain an injection molding part positioning signal; and the control module controls the detection module to perform magnetic detection on the injection molding part to be detected according to the injection molding part positioning signal to generate a corresponding injection molding part screw detection result. In the embodiment of the present application, the upper computer receives the screw hole image of the injection molding part to be detected, the image is denoised, enhanced and analyzed for features by the preset deep learning model, the installation position of the screw in the deep hole is identified, and data basis is provided for subsequent mechanical positioning; the control module receives the screw hole coordinate information sent by the upper computer to move the injection molding part to be detected, and generates positioning information to ensure that the detection module works after the injection molding part is stable, and the data reliability is improved. The detection module scans the screw hole area after the injection molding part is positioned, and judges whether the screw is missed by detecting the magnetic signal of the screw, so as to combine the deep learning positioning and the magnetic signal analysis to eliminate the subjective omissions of manual detection. Through the cooperative control of the upper computer, the control module and the detection module, manual intervention is not needed from image input to result output, human cost is saved, different injection molding part models and deep hole positions can be adapted, the general problem of traditional automatic detection is solved, and therefore all the screw installation conditions of the injection molding parts can be efficiently detected. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without any creative effort.

[0010] Figure 1 A flowchart of a screw detection method for injection molding parts based on a deep learning model is provided for the embodiment of the present application.

[0011] Figure 2 A sub-flow schematic diagram of the injection molding part screw detection method based on the deep learning model is provided for the embodiment of the present application.

[0012] Figure 3 A sub-flow schematic diagram of the injection molding part screw detection method based on the deep learning model is provided for the embodiment of the present application.

[0013] Figure 4 A sub-flow schematic diagram of the injection molding part screw detection method based on the deep learning model is provided for the embodiment of the present application.

[0014] Figure 5 A sub-flow schematic diagram of the injection molding part screw detection method based on the deep learning model is provided for the embodiment of the present application.

[0015] Figure 6 A sub-flow schematic diagram of the injection molding part screw detection method based on the deep learning model is provided for the embodiment of the present application.

[0016] Figure 7 A schematic block diagram of the injection molding part screw detection system based on the deep learning model is provided for the embodiment of the present application.

[0017] Figure 8 A schematic block diagram of the computer device is provided for the embodiment of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0019] It should be understood that, when used in the present specification and the appended claims, the terms “comprise” and “include” indicate the presence of the described features, integers, steps, operations, elements, and / or components, but do not exclude one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0020] It should also be understood that the terms used in the present specification and the appended claims are only for the purpose of describing particular embodiments and do not intend to limit the present application. As used in the present specification and the appended claims, the singular forms “a”, “an” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0021] It should be further understood that the term "and / or" used in the description and claims of the application herein is used to mean any one and / or any combination of the associated listed items and includes all possible combinations.

[0022] Referring to Figure 1 , Figure 1 A flowchart of a deep learning model-based injection molding part screw detection method provided by an embodiment of the present application is shown. The deep learning model-based injection molding part screw detection method in this embodiment can be applied to key processes in the intelligent production of air conditioners, such as air conditioner outdoor unit sheet metal assembly, indoor unit plastic shell fastening, and air conditioner pipeline system connector detection. The deep learning model-based injection molding part screw detection method is applied to the host computer, the control module, and the detection module. The programmable logic controller in the control module can control the air cylinder to move the magnetic proximity switch in the detection module to a position where magnetic detection can be performed, thereby performing screw detection of the injection molding part. This method can efficiently detect screws in various injection molding parts.

[0023] Figure 1 A flowchart of a deep learning model-based injection molding part screw detection method provided by an embodiment of the present application is shown. As shown in the figure, the method includes the following steps S110-S130.

[0024] S110, the host computer inputs the received screw hole image of the injection molding part to be detected into a preset deep learning model to generate screw hole coordinate information to be detected and sends it to the control module.

[0025] In this embodiment, the host computer is a computer responsible for advanced calculation and decision-making in an industrial control system. In an air conditioner production line, the host computer can communicate with an industrial camera and a PLC controller in real time through a gigabit Ethernet. The screw hole image is an image captured by a high-resolution industrial camera. The preset deep learning model is a trained deep learning model, such as a YOLOv7 target detection framework. First, an industrial camera is used to take a picture of the injection molding part to obtain the screw hole position image of the injection molding part, which is uploaded to the host computer. The host computer receives the screw hole image of the injection molding part to be detected and inputs it into the preset deep learning model. The trained model can directly generate the screw hole coordinates to be detected after inference, such as (1, 1), (1, 2), and (2, 2). The host computer can send the generated screw hole coordinate information to the control module through a preset communication method (such as gigabit Ethernet). The preset deep learning model generates the screw hole coordinate information to be detected, so that the subsequent detection of the position where the screw should exist can be accurately performed.

[0026] In an embodiment, asFigure 2 As shown, the step S110 further includes steps S1101-S1102.

[0027] S1101, constructing an initial deep learning model according to a deep learning framework;

[0028] S1102, training the initial deep learning model according to the collected training screw hole position image and the label information of the image to obtain the preset deep learning model.

[0029] In the embodiment, the deep learning framework is an open-source deep learning library based on Python language, such as TensorFlow (including Keras interface), PyTorch, etc., which provides a full-process tool for model construction, training and deployment. The initial deep learning model is an untrained neural network structure, which includes convolutional layers, pooling layers, fully connected layers, etc., and needs to learn parameters through data. The label information of the image is the annotated coordinate data. According to the deep learning framework, the initial deep learning model is constructed, for example, a convolutional layer is constructed by using 9 convolutional kernels with a size of 5x5, a ReLU activation function is introduced after each convolutional layer (a total of 7 places, which needs to be adjusted according to the actual network depth, and it is recommended to match the number of convolutional layers here), and the non-linear expression ability is enhanced; two kinds of pooling layers are constructed in parallel: 3 Average Pooling layers (used to retain background features) and 1 Max Pooling layer (used to highlight significant features); the output tensors of different pooling layers are unified to the same dimension (such as 128 dimensions) through the dim=1 parameter, and then weighted fusion is performed through the Sigmoid function to generate an attention weight map; the fused feature map is flattened into a one-dimensional vector (Flatten operation); 2 fully connected layers (containing 256 / 128 neurons respectively) are connected to reduce the dimension of the features; finally, the regression layer (output dimension is 2, corresponding to the X / Y coordinates of the screw hole) outputs the prediction result, wherein the specific construction process is not limited, and any reasonable initial deep learning model can be generated. The training screw hole position image containing different sizes, shapes and surface features of the injection molding screw hole position, and the label information of the image are used to train the convolutional and attention mechanism hybrid neural network to obtain a preset deep learning model that can accurately generate screw hole position coordinates. By constructing and training the deep learning model, accurate screw hole position coordinate information is obtained, which facilitates subsequent accurate detection.

[0030] In an embodiment, as shown in Figure 3 The step S1102 further includes steps S11021-S11022.

[0031] S11021, training the convolutional and attention mechanism hybrid neural network of the initial deep learning model according to the training screw hole position image and the label information;

[0032] S11022, evaluate the training result by a preset cross loss function, and update the model parameters by a preset optimization algorithm according to the evaluation result, to generate the preset deep learning model.

[0033] In the embodiment, the training screw hole position image is the collected image data containing the screw hole, and the label information is the artificial annotation data in supervised learning, which is the accurate coordinates of the screw hole in the image. The convolution and attention mechanism mixed neural network of the initial deep learning model is trained according to the training screw hole position image and the label information. Specifically, first, collect screw hole position image samples of injection molding parts with different sizes. These images should contain different sizes, shapes, and surface features of the injection molding part screw hole position. The collected training screw hole position image is labeled with the corresponding two-dimensional coordinate label information, such as (1, 1), (2, 2), etc. In order to increase the diversity and reliability of the training samples, data enhancement processing can be performed. For example, the data set is expanded by rotating, flipping, scaling, and translating the image, etc. The convolution and attention mechanism mixed neural network of the initial deep learning model is trained. Specifically, the attention mechanism is introduced to improve the weight of the key features of the screw hole position in the model. During the training process, the attention weight is dynamically adjusted according to the importance of the image features, so that the model can focus on processing the most relevant features, and the convolution parameters of the convolution layer are trained and adjusted. During the model training, the classification accuracy of the model is measured by a preset cross-entropy loss function (cross-entropy loss), and the model parameters are updated by a preset optimization algorithm. Specifically, the model parameters can be optimized by a gradient descent algorithm. The training and updating are terminated when the preset number of iterations is reached, the validation set performance is saturated, or early stopping (Early Stopping) is performed. The optimal parameters (such as the weights when the validation set loss is the lowest) are saved as the final model, and are used as the preset deep learning model. The initial model is trained and updated to generate a deep learning model that can efficiently and accurately locate the screw hole position.

[0034] In an embodiment, as shown in FIG. 1 1, the step S110 further includes steps S111-S113. Figure 4

[0035] S111, extracting image features of the screw hole image by the preset deep learning model;

[0036] S112, generating attention features of the target region according to the dot product of the preset attention weight and the image features;

[0037] S113, generating the to-be-detected screw hole coordinate information by the regression network of the preset deep learning model from the attention features.​

[0038] In the embodiment, the image feature refers to an attribute or pattern that can reflect the essence and important information of the image. In the screw hole image, the image feature can include the shape, size, edge texture, color distribution, etc. of the screw hole. The attention feature is a feature vector obtained by the dot product operation of the preset attention weight and the image feature. It fuses the information of the original image feature and the attention weight, and highlights the features related to the screw hole detection, which helps the model to better identify the position and shape of the screw hole. The image feature of the screw hole image is extracted by the preset deep learning model. Specifically, the screw hole image is input into the preset deep learning model. The convolution layer, pooling layer, etc. in the model will perform layer-by-layer feature extraction on the image. The convolution layer extracts local features such as edges and textures by sliding the convolution kernel on the image; the pooling layer down-samples the feature map to reduce the data volume while enhancing the robustness of the features. After multiple layers of processing, the model obtains an image feature vector that can reflect the essential features of the screw hole image. The extracted image feature vector is subjected to dot product operation with the preset attention weight vector. Each element in the attention weight vector corresponds to a feature dimension or a region in the image feature vector. Through the dot product operation, the image features are weighted, so that the feature dimensions or regions with high relevance to the screw hole detection will get larger weights, and thus will be more prominent in the attention feature, while the features with low relevance will be suppressed. The attention feature vector is input into the regression network of the preset deep learning model. The fully connected layer in the regression network performs nonlinear transformation and fitting on the attention feature, gradually mapping the feature to the output space of the screw hole coordinate information. The screw hole coordinate information to be detected is generated according to the attention feature, so as to ensure the accuracy of the coordinate information.

[0039] In an embodiment, the step S110 further comprises a step S114.

[0040] S114, sending the screw hole coordinate information to be detected to the control module through the 6G communication module.

[0041] In the embodiment, the 6G communication module is the sixth generation mobile communication technology, which adopts advanced modulation and coding technology, massive antenna array (Massive MIMO), etc., to improve the transmission rate and reliability of the signal. The 6G communication module sends the to-be-detected screw hole coordinate information to the control module. Specifically, the to-be-detected screw hole coordinate information is converted into a digital signal suitable for transmission in a wireless channel. Then the digital signal is modulated, up-converted, etc. by the 6G millimeter wave communication device, loaded onto the millimeter wave carrier, and finally transmitted by the antenna. After the receiver at the control module end receives the signal, it is down-converted, demodulated, etc. to recover the original coordinate information. It can be understood that other processes requiring data transmission in the embodiment also use the 6G communication module. The information is transmitted by the 6G communication module, so that the entire detection link of the injection molding screw is completed in an efficient transmission environment, realizing long-distance transmission and real-time uploading of data.

[0042] S120, the control module moves the to-be-detected injection molding part to the target position according to the to-be-detected screw hole coordinate information to obtain an injection molding part in-place signal.

[0043] In the embodiment, the control module is a module composed of a programmable logic controller (PLC). PLC is a digital operation electronic system specially designed for industrial environment. According to the to-be-detected screw hole coordinate information, the to-be-detected injection molding part is moved to the target position to obtain an injection molding part in-place signal. Specifically, the PLC controls the conveying device to move the to-be-detected injection molding part to the target position specified by the coordinate information, ensuring that the injection molding screw hole is aligned with the screw detection device. When the to-be-detected injection molding part touches the injection molding part in-place sensor in the screw detection device, the in-place sensor sends a signal to generate an injection molding part in-place signal to the PLC. At this time, the control module obtains the injection molding part in-place signal. By moving the to-be-detected injection molding part to the target position to obtain the injection molding part in-place signal, it is ensured that the to-be-detected injection molding part reaches the specified position.

[0044] In an embodiment, as shown in Figure 5 The step S120 further includes steps S121-S122.

[0045] S121, calculating the movement information of the to-be-detected injection molding part according to the to-be-detected screw hole coordinate information;

[0046] S122, moving the to-be-detected injection molding part according to the movement information, and generating an injection molding part in-place signal if it moves to the target position.

[0047] In the embodiment, the movement information is a set of parameters for guiding the movement of the injection molded part to be detected, which can include the direction and distance of the movement of the injection molded part. The control module further includes a conveying device, which is a device for moving the injection molded part to be detected from one position to another. Common conveying devices include conveyors, mechanical arms, sliding rails, etc. The target position is a pre-set accurate position that the injection molded part needs to reach according to the detection requirements. At this position, the screw hole of the injection molded part can be accurately aligned with the screw detection device, so as to facilitate the subsequent detection of the screw hole. In the embodiment, the target position can be the position where the coordinates of the screw hole to be detected are located. The injection molded part in position signal is a signal for indicating that the injection molded part to be detected has been successfully moved to the target position. The movement information of the injection molded part to be detected is calculated according to the coordinate information of the screw hole to be detected. Specifically, the relevant parameters of how the injection molded part to be detected needs to move are derived through specific algorithms or logical operations according to the known coordinate information of the screw hole to be detected. These parameters combined together are the movement information. For example, in two dimensions, the screw hole has a specific coordinate position (e.g. represented by (x, y) in a two-dimensional plane), and the injection molded part also has an initial position coordinate. By calculating the difference between the coordinate of the screw hole and the initial position coordinate of the injection molded part, and combining the movement direction and mode of the conveying device (such as linear motion, rotary motion, etc.), it can be determined that the injection molded part needs to move along which axes, the distance of the movement, whether it needs to rotate, and the angle of rotation, etc. These determined data about how the injection molded part moves constitute the movement information. The injection molded part to be detected is moved according to the movement information, and if it is moved to the target position, the injection molded part in position signal is generated. Specifically, the PLC (Programmable Logic Controller) will control the conveying device (such as a motor-driven conveyor, a mechanical arm, etc.) to start moving according to the movement information, so that the injection molded part to be detected moves in the manner specified by the movement information. During the movement, the system will monitor the position state of the injection molded part in real time. When the injection molded part is accurately moved to the pre-set target position (i.e. the position where the screw hole can be accurately aligned with the screw detection device), the injection molded part to be detected will touch the injection molded part in position sensor in the screw detection device, and the in position sensor will send the injection molded part in position signal to the PLC, informing it that the injection molded part has reached the specified position, and the subsequent operation can be carried out according to the pre-determined program, such as controlling the screw detection device to start working, etc. The movement information of the injection molded part is accurately calculated through the coordinate information, and the injection molded part is moved according to the movement information and the in position signal is generated, which provides a reliable basis for accurately moving the injection molded part and ensures that the injection molded part is accurately positioned, triggering the orderly development of the subsequent detection process.

[0048] S130, the control module controls the detection module to perform magnetic detection on the injection molded part to be detected according to the injection molded part in position signal, to generate a corresponding injection molded part screw detection result.

[0049] In the embodiment, the detection module is a component for performing a specific detection task, which is mainly used for magnetic detection of the screw of the injection molded part in the scenario. It can include magnetic sensors (such as magnetic proximity switches, etc.), which can sense the magnetic characteristics of the screw. The injection molded part screw detection result is the final report on the state of the screw generated after the magnetic detection of the injection molded part screw. This result is presented in various forms, such as by indicating light to show whether the detection is qualified (green light indicates qualified, red light indicates unqualified), or in the form of a digital signal transmitted to the control module, which then uploads the result to the upper computer for display and storage. It can also contain detailed detection data such as the magnetic strength of the screw, position deviation, etc., so that the operator can accurately evaluate the quality of the injection molded part. The control module controls the detection module to perform magnetic detection on the injection molded part to be detected according to the injection molded part in-place signal to generate the corresponding injection molded part screw detection result. Specifically, when the control module receives the injection molded part in-place signal, the control module controls the cylinder in the detection module to move the screw detection magnetic proximity switch downward, and when the magnetic proximity switch touches the magnetic proximity switch in-place sensor, the sensor sends a signal to the PLC, which controls the cylinder to stop the movement of the screw detection magnetic proximity switch. The screw detection magnetic proximity switch in the detection module detects the magnetic detection of the injection molded part to be detected after the screw detection magnetic proximity switch is in place to generate the corresponding injection molded part screw detection result. For example, the screw detection magnetic proximity switch judges whether there is a screw in the screw hole by whether it senses magnetism. If there is magnetism, the indicator light turns green, indicating that there is a screw in the screw hole, and the injection molded part screw detection result is generated. By detecting the magnetic detection of the injection molded part to be detected by the detection module to generate the corresponding injection molded part screw detection result, the precision detection of the screw in the injection molded part is realized automatically, and the screw detection efficiency is improved.

[0050] In an embodiment, as shown in Figure 6 The step S130 further includes steps S131-S33 before it.

[0051] S131, judge whether the magnetic of the screw is detected at the coordinate information of the screw hole to be detected;

[0052] S132, if the magnetic of the screw is not detected, generate a detection signal of magnetic detection failure and generate an alarm information;

[0053] S133, generate the injection molded part screw detection result according to the magnetic detection failure detection signal and the alarm information.

[0054] In the embodiment, the detection module performs magnetic detection on the to-be-detected injection molded part to determine whether the magnetic property of a screw is detected at the to-be-detected screw hole coordinate information. Specifically, the screw detection magnetic proximity switch in the detection module detects whether the magnetic property exists at the to-be-detected screw hole coordinate information. If the magnetic property exists, the indicator light turns green, that is, the injection molded part screw detection result that a screw exists in the screw hole is generated. If the magnetic property of the screw is not detected, the detection signal that the magnetic detection fails is generated, for example, the indicator light turns red and the buzzer alarm is controlled, indicating that no screw exists in the screw hole, the injection molded part screw detection result that the detection is unqualified and no screw exists in the screw hole is generated. Meanwhile, the screw detection magnetic proximity switch in the detection module sends a signal to the PLC (control module), the PLC controls the air cylinder to reset the screw detection magnetic proximity switch, thereby completing the detection of the injection molded part screw and preparing for the next injection molded part detection. The magnetic detection is performed at the preset to-be-detected screw hole coordinate position, whether an abnormality exists is determined according to the detection result, and then corresponding signals and information are generated, and finally the complete injection molded part screw detection result is formed, so as to guarantee the detection accuracy of the screw installation quality of the injection molded part.

[0055] Figure 7 is a schematic block diagram of an injection molded part screw detection system 200 based on a deep learning model provided by an embodiment of the present application. As shown in Figure 7 corresponding to the above injection molded part screw detection method based on a deep learning model, the present application also provides an injection molded part screw detection system based on a deep learning model. The injection molded part screw detection system based on a deep learning model includes units for executing the above injection molded part screw detection method based on a deep learning model. The system can be configured in a terminal such as a desktop computer, a tablet computer, a laptop computer, etc. Specifically, please refer to Figure 7 , the injection molded part screw detection system based on a deep learning model includes a host computer 210, a control module 220, and a detection module 232.

[0056] The host computer 210 is configured to input the received screw hole image of the to-be-detected injection molded part into a preset deep learning model to generate to-be-detected screw hole coordinate information and send it to the control module.

[0057] In an embodiment, the host computer 210 includes a configuration unit and a training unit.

[0058] The construction unit is configured to construct an initial deep learning model according to a deep learning framework.

[0059] The training unit is configured to train the initial deep learning model according to the collected training screw hole position image and the label information of the image to obtain the preset deep learning model.

[0060] In an embodiment, the host computer 210 includes an attention training unit and an optimization updating unit.

[0061] An attention training unit is used to train the initial deep learning model's convolutional and attention mechanism hybrid neural network based on the training screw hole position image and the label information;

[0062] The optimization and update unit is used to evaluate the training results through a preset cross-loss function and update the model parameters according to the evaluation results through a preset optimization algorithm to generate the preset deep learning model.

[0063] In one embodiment, the host computer 210 includes a feature extraction unit, a feature generation unit, and a coordinate generation unit.

[0064] The feature extraction unit is used to extract image features of the screw hole image through the preset deep learning model;

[0065] The feature generation unit is used to generate attention features of the target region based on the dot product of the preset attention weights and the image features;

[0066] The coordinate generation unit is used to generate the coordinate information of the screw hole to be detected by passing the attention features through the regression network of the preset deep learning model.

[0067] In one embodiment, the host computer 210 includes an information transmission unit.

[0068] The information transmission unit is used to send the coordinate information of the screw hole to be detected to the control module via the 6G communication module.

[0069] The control module 220 is used to move the injection molded part to be tested to the target position according to the coordinate information of the screw hole to be tested to obtain the injection molded part positioning signal.

[0070] In one embodiment, the host computer 210 includes an information computing unit and a signal generation unit.

[0071] An information calculation unit is used to calculate the movement information of the injection molded part to be tested based on the coordinate information of the screw hole to be tested;

[0072] The signal generation unit is used to move the injection molded part to be detected according to the movement information, and generate an injection molded part arrival signal if it moves to the target position.

[0073] The control module 230 is used to control the detection module to perform magnetic detection on the injection molded part to be detected according to the injection molded part positioning signal, so as to generate the corresponding injection molded part screw detection result.

[0074] In one embodiment, the control module 230 includes a magnetic determination unit, an alarm generation unit, and a detection report generation unit.

[0075] a magnetic determination unit configured to determine whether a magnetic property of a screw is detected at the coordinate information of the screw hole to be detected;

[0076] an alarm generation unit configured to generate a detection signal of a magnetic property detection failure and generate alarm information if the magnetic property of the screw is not detected;

[0077] a detection report generation unit configured to generate the injection molding part screw detection result according to the detection signal of the magnetic property detection failure and the alarm information.

[0078] It should be noted that the specific implementation process of the injection molding part screw detection system 200 and each unit based on the deep learning model can be clearly understood by those skilled in the art, and can refer to the corresponding description in the foregoing method embodiments. For the convenience and brevity of description, it will not be repeated here.

[0079] The injection molding part screw detection system based on the deep learning model can be implemented in the form of a computer program, which can run on a computer device as shown in Figure 8 .

[0080] Please refer to Figure 8 , Figure 8 is a schematic block diagram of a computer device provided by an embodiment of the present application. The computer device 500 can be a terminal or a server, wherein the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a personal digital assistant, a wearable device, and an electronic device with a communication function. The server can be a stand-alone server or a server cluster composed of multiple servers.

[0081] Please refer to Figure 8 , the computer device 500 includes a processor 502, a memory, and a network interface 505 connected through a system bus 501, wherein the memory can include a non-volatile storage medium 503 and an internal memory 504.

[0082] The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. The computer program 5032 includes program instructions, which when executed, can cause the processor 502 to perform a composite laser printing method.

[0083] The processor 502 is configured to provide computing and control capabilities to support the operation of the entire computer device 500.

[0084] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503, which when executed by the processor 502, can cause the processor 502 to perform an injection molding part screw detection method based on a deep learning model.

[0085] The network interface 505 is configured to perform network communication with other devices. Those skilled in the art can understand that the network interface 505 can be configured to perform network communication through wired or wireless communication technology. Figure 8 The structure shown in FIG. 5 is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device 500 to which the scheme of the present application is applied. Specifically, the computer device 500 can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0086] The processor 502 is configured to run the computer program 5032 stored in the memory, so as to implement the steps of the above method.

[0087] It should be understood that, in the embodiments of the present application, the processor 502 can be a central processing unit (CPU), and the processor 502 can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0088] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments of the method can be completed by a computer program instructing related hardware. The computer program includes program instructions, and the computer program can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the above-mentioned embodiments of the method.

[0089] Therefore, the present application also provides a storage medium. The storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions. The program instructions are executed by the processor to make the processor perform the steps of the above method.

[0090] The storage medium can be a U disk, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk, and various computer-readable storage media that can store program codes.

[0091] Those skilled in the art can understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been described in general terms in the above description. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0092] In several embodiments provided by the present application, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiments described above are only illustrative. For example, the division of each unit is only a logical functional division, and actual implementation can have another division. For example, multiple units or components can be combined or integrated into another system, or some features can be omitted or not executed.

[0093] The steps in the method embodiments of the present application can be adjusted, combined and deleted in sequence according to actual needs. The units in the system embodiments of the present application can be combined, divided and deleted according to actual needs. In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit.

[0094] The integrated unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art, or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a terminal or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application.

[0095] The above description is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the present application, and these modifications or replacements should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for detecting injection molded part screws based on a deep learning model, characterized in that, The method is applied to an upper computer, a control module and a detection module, and comprises the following steps: The upper computer inputs a received screw hole image of a to-be-detected injection molding part into a preset deep learning model to generate to-be-detected screw hole coordinate information and sends the to-be-detected screw hole coordinate information to the control module; The control module moves the to-be-detected injection molding part to a target position according to the to-be-detected screw hole coordinate information to obtain an injection molding part in-position signal; The control module controls the detection module to perform magnetic detection on the to-be-detected injection molding part according to the injection molding part in-position signal to generate a corresponding injection molding part screw detection result.

2. The method of claim 1, wherein, Before the step of inputting the received screw hole image of the to-be-detected injection molding part into the preset deep learning model to generate the to-be-detected screw hole coordinate information, the method comprises the following steps: An initial deep learning model is constructed according to a deep learning framework; The initial deep learning model is trained according to collected training screw hole position images and label information of the images to obtain the preset deep learning model.

3. The method of claim 2, wherein, The step of training the initial deep learning model according to the collected training screw hole position images and the label information of the images to obtain the preset deep learning model comprises the following steps: The convolution and attention mechanism mixed neural network of the initial deep learning model is trained according to the training screw hole position images and the label information; A training result is evaluated through a preset cross-loss function, and model parameters are updated through a preset optimization algorithm according to the evaluation result to generate the preset deep learning model.

4. The method of claim 1, wherein, The step of inputting the received screw hole image of the to-be-detected injection molding part into the preset deep learning model to generate the to-be-detected screw hole coordinate information comprises the following steps: Image features of the screw hole image are extracted through the preset deep learning model; Attention features of a target region are generated according to a preset attention weight and a dot product of the image features; The attention features are input into a regression network of the preset deep learning model to generate the to-be-detected screw hole coordinate information.

5. The method of claim 1, wherein, The step of generating the to-be-detected screw hole coordinate information and sending the to-be-detected screw hole coordinate information to the control module comprises the following steps: The to-be-detected screw hole coordinate information is sent to the control module through a 6G communication module.

6. The method of claim 1, wherein, The step of moving the to-be-detected injection molding part to the target position according to the to-be-detected screw hole coordinate information to obtain the injection molding part in-position signal comprises the following steps: Movement information of the to-be-detected injection molding part is calculated according to the to-be-detected screw hole coordinate information; The to-be-detected injection molding part is moved according to the movement information, and an injection molding part in-position signal is generated if the to-be-detected injection molding part is moved to the target position.

7. The method of claim 1, wherein, The step of performing magnetic detection on the to-be-detected injection molding part to generate the corresponding injection molding part screw detection result comprises the following steps: It is judged whether the magnetic property of a screw is detected at the to-be-detected screw hole coordinate information; If the magnetic property of the screw is not detected, a magnetic detection failure detection signal is generated, and an alarm information is generated; The injection molding part screw detection result is generated according to the magnetic detection failure detection signal and the alarm information.

8. A deep learning model-based injection molded part screw detection system, characterized by, The system comprises an upper computer, a control module and a detection module, wherein: The host computer is configured to input a received screw hole image of an injection molded part to be detected into a preset deep learning model to generate screw hole coordinate information to be detected and send the screw hole coordinate information to the control module; The control module is configured to move the injection molded part to be detected to a target position according to the screw hole coordinate information to be detected to obtain an injection molded part in-position signal; The control module is configured to control the detection module to perform magnetic detection on the injection molded part to be detected according to the injection molded part in-position signal to generate a corresponding injection molded part screw detection result.

9. A deep learning model-based injection molded part screw detection system, characterized by, The computer program product comprises at least three computer devices, each of which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processors of the at least three computer devices collectively implement the method of any one of claims 1 to 7 when executing the corresponding computer programs.

10. A storage medium, characterized by The storage medium stores a computer program, and the computer program comprises program instructions executable by the processor to implement the method of any one of claims 1 to 7.

Images