A die-cutting processing device and method with high profiling sharpness and integrated waste discharge

Abstract

The present application relates to the field of intelligent processing, and specifically discloses a die-cutting processing device and method with high profiling sharpness and integrated waste discharge, which transmits the surface state image of the die-cutting processing formed by the industrial camera deployed in the die-cutting processing device to the molding quality monitoring server of the die-cutting processing device, and uses AI-based image recognition and processing technology in the molding quality monitoring server to perform super-pixel segmentation and encoding on the die-cutting surface state image, so as to automatically judge whether the molding quality of the die-cutting processing formed meets the predetermined standard according to the fusion semantic representation features of each super-pixel block in the whole area. In this way, the molding quality monitoring server can be used to quickly analyze the surface state of the die-cutting device, which greatly improves the detection speed and the accuracy of quality control compared with manual inspection.

Description

Technical Field

[0001] The present application relates to the field of intelligent processing, and more specifically, to a die-cutting processing device and a processing method with high profiling sharpness and integrated waste discharge. Background Art

[0002] With the improvement of quality requirements for industrial products, the processing accuracy of die cutters directly affects the quality of the final products. Therefore, high-precision die cutters are needed to meet this demand. The die cutter processing equipment with high profiling sharpness and integrated waste removal is a high-precision equipment used in a die-cutting process. It can remove waste while cutting materials.

[0003] During the die cutting process, the molding quality of the die directly affects the dimensional accuracy and appearance quality of the die-cut products. Therefore, through testing, potential problems of the die cutting, such as dimensional deviation and shape error, can be discovered in time to avoid the production of unqualified products.

[0004] However, the molding quality inspection of traditional die cutting relies heavily on the experience and skills of the operator, and it is difficult for the human eye to detect tiny defects, which makes it difficult to ensure the consistency and accuracy of the inspection results, thus affecting the performance and reliability of the die cutting.

[0005] Therefore, an optimized die-cutting processing equipment with high profiling sharpness and integrated waste discharge is desired. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a knife die processing equipment and processing method with high profiling sharpness and integrated waste discharge, which transmits the surface state image of the processed knife die collected by the industrial camera deployed in the knife die processing equipment to the molding quality monitoring server of the knife die processing equipment, and uses AI-based image recognition and processing technology in the molding quality monitoring server to perform super-pixel segmentation and encoding on the knife die surface state image, so as to automatically judge whether the molding quality of the processed knife die meets the predetermined standard according to the fusion semantic representation features of each super-pixel block in the whole area. In this way, the molding quality monitoring server can be used to quickly analyze the surface state of the knife die, which greatly improves the detection speed and quality control accuracy compared with manual inspection.

[0007] According to one aspect of the present application, a die-cutting processing device with high profiling sharpness and integrated waste discharge is provided, comprising:

[0008] The die cutting surface state image acquisition module is used to acquire the surface state image of the die cutting mold through the industrial camera deployed on the die cutting processing equipment;

[0009] An image transmission module, used for transmitting the surface state image of the cutting die to a molding quality monitoring server of the cutting die processing equipment;

[0010] A superpixel segmentation module is used to perform superpixel segmentation on the surface state image of the knife die on the molding quality monitoring server to obtain a sequence of superpixel blocks of the surface state of the knife die;

[0011] A superpixel block context semantic coding module is used to perform superpixel block context semantic coding on the sequence of the die surface state superpixel blocks on the molding quality monitoring server to obtain a sequence of context die surface state superpixel block semantic coding feature vectors;

[0012] A context semantic enhancement module is used to input the sequence of context die surface state super pixel block semantic encoding feature vectors into the context semantic enhancement module based on saliency-globality in the molding quality monitoring server to obtain a sequence of context enhanced die surface state super pixel block semantic encoding feature vectors;

[0013] A global fusion module is used for inputting the sequence of context-enhanced superpixel semantic encoding feature vectors of the surface state of the die into the energy measurement attention fusion network based on sequence annotation on the molding quality monitoring server to obtain a global fusion feature vector of the surface state of the die as a global fusion feature of the surface state of the die;

[0014] The quality inspection result generating module is used to obtain the quality inspection result based on the global fusion feature of the surface state of the die cutter in the molding quality monitoring server, and the quality inspection result is used to indicate whether the molding quality of the die cutter used for molding meets the predetermined standard.

[0015] According to another aspect of the present application, a cutting die processing method with high profiling sharpness and integrated waste discharge is provided, which comprises:

[0016] The surface state image of the processed die is collected by an industrial camera deployed in the die processing equipment;

[0017] Transmitting the surface state image of the cutting die to a molding quality monitoring server of the cutting die processing equipment;

[0018] On the molding quality monitoring server, super-pixel segmentation is performed on the surface state image of the knife die to obtain a sequence of super-pixel blocks of the surface state of the knife die;

[0019] On the molding quality monitoring server, the sequence of super pixel blocks of the die surface state is subjected to super pixel block context semantic coding to obtain a sequence of context die surface state super pixel block semantic coding feature vectors;

[0020] In the molding quality monitoring server, the sequence of contextual die surface state super pixel block semantic coding feature vectors is input into a contextual semantic enhancement module based on saliency-globality to obtain a sequence of contextual enhanced die surface state super pixel block semantic coding feature vectors;

[0021] In the molding quality monitoring server, the sequence of context-enhanced superpixel semantic encoding feature vectors of the die surface state is input into an energy-metric attention fusion network based on sequence annotation to obtain a die surface state global fusion feature vector as a die surface state global fusion feature;

[0022] In the molding quality monitoring server, a quality inspection result is obtained based on the global fusion feature of the surface state of the cutting die, and the quality inspection result is used to indicate whether the molding quality of the processed cutting die meets the predetermined standard.

[0023] Compared with the prior art, the present application provides a die-cutting processing equipment and method with high profiling sharpness and integrated waste discharge, which transmits the surface state image of the die-cutting processing formed by the industrial camera deployed in the die-cutting processing equipment to the forming quality monitoring server of the die-cutting processing equipment, and uses AI-based image recognition and processing technology in the forming quality monitoring server to perform super-pixel segmentation and encoding on the die-cutting surface state image, so as to automatically judge whether the forming quality of the die-cutting processing formed meets the predetermined standard according to the fusion semantic representation features of each super-pixel block in the whole area. In this way, the forming quality monitoring server can be used to quickly analyze the surface state of the die-cutting, which greatly improves the detection speed and quality control accuracy compared with manual inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0025] Figure 1 It is a block diagram of a die-cutting processing device with high profiling sharpness and integrated waste discharge according to an embodiment of the present application.

[0026] Figure 2 It is a schematic diagram of the structure of a die-cutting processing device with high profiling sharpness and integrated waste discharge according to an embodiment of the present application.

[0027] Figure 3 It is a block diagram of a training module in a die-cutting processing device with high profiling sharpness and integrated waste discharge according to an embodiment of the present application.

[0028] Figure 4 It is a flow chart of a cutting die processing method with high profiling sharpness and integrated waste discharge according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] The embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, but rather these embodiments are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0030] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0031] In the description of the embodiments of the present disclosure, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0032] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0033] With the improvement of quality requirements for industrial products, the processing accuracy of die cutters directly affects the quality of the final products. Therefore, high-precision die cutters are needed to meet this demand. The die cutter processing equipment with high profiling sharpness and integrated waste removal is a high-precision equipment used in a die-cutting process. It can remove waste while cutting materials.

[0034] During the die cutting process, the molding quality of the die directly affects the dimensional accuracy and appearance quality of the die-cut products. Therefore, through testing, potential problems of the die cutting, such as dimensional deviation and shape error, can be discovered in time to avoid the production of unqualified products.

[0035] However, the molding quality inspection of traditional die cutting relies heavily on the experience and skills of the operator, and it is difficult for the human eye to detect tiny defects, which makes it difficult to ensure the consistency and accuracy of the inspection results, thus affecting the performance and reliability of the die cutting.

[0036] Therefore, in response to the above technical problems, the technical concept of the present application is to transmit the surface state image of the processed die collected by the industrial camera deployed in the die processing equipment to the molding quality monitoring server of the die processing equipment, and use AI-based image recognition and processing technology in the molding quality monitoring server to perform super-pixel segmentation and encoding on the die surface state image, so as to automatically judge whether the molding quality of the processed die meets the predetermined standard according to the fusion semantic representation features of each super-pixel block in the whole area. In this way, the molding quality monitoring server can be used to quickly analyze the surface state of the die, which greatly improves the detection speed and quality control accuracy compared with manual inspection.

[0037] Figure 1 It is a block diagram of a die-cutting processing device with high profiling sharpness and integrated waste discharge according to an embodiment of the present application. Figure 2 Schematic diagram of the structure of the die-cutting processing equipment with high profiling sharpness and integrated waste discharge according to the embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the knife die processing equipment 100 with high profiling sharpness and integrated waste discharge includes: a knife die surface state image acquisition module 110, which is used to acquire the surface state image of the knife die processed by an industrial camera deployed in the knife die processing equipment; an image transmission module 120, which is used to transmit the surface state image of the knife die to the molding quality monitoring server of the knife die processing equipment; a superpixel segmentation module 130, which is used to perform superpixel segmentation on the surface state image of the knife die on the molding quality monitoring server to obtain a sequence of knife die surface state superpixel blocks; a superpixel block context semantic encoding module 140, which is used to perform superpixel block context semantic encoding on the sequence of knife die surface state superpixel blocks on the molding quality monitoring server to obtain a sequence of context knife die surface state superpixel block semantic encoding feature vectors; context semantic The enhancement module 150 is used to input the sequence of contextual knife die surface state super pixel block semantic coding feature vectors into the contextual semantic enhancement module based on significance-globality in the molding quality monitoring server to obtain the sequence of contextual enhanced knife die surface state super pixel block semantic coding feature vectors; the global fusion module 160 is used to input the sequence of contextual enhanced knife die surface state super pixel block semantic coding feature vectors into the energy measurement attention fusion network based on sequence annotation in the molding quality monitoring server to obtain the knife die surface state global fusion feature vector as the knife die surface state global fusion feature; and the quality inspection result generation module 170 is used to obtain the quality inspection result based on the global fusion feature of the knife die surface state in the molding quality monitoring server, and the quality inspection result is used to indicate whether the molding quality of the processed knife die meets the predetermined standard.

[0038] In the embodiment of the present application, the die surface state image acquisition module 110 is used to acquire the surface state image of the die after processing by an industrial camera deployed in the die processing equipment. It should be understood that the surface state image of the die can reflect various characteristics of the die surface, such as roughness, texture, color, shape, and possible defects. Based on this, in the technical solution of the present application, the die surface state image of the die after processing is acquired by an industrial camera deployed in the die processing equipment and image analysis and feature extraction are performed to ensure that the die meets the quality requirements.

[0039] In the embodiment of the present application, the image transmission module 120 is used to transmit the surface state image of the cutting die to the molding quality monitoring server of the cutting die processing equipment. Accordingly, by transmitting the surface state image of the cutting die to the molding quality monitoring server of the cutting die processing equipment, real-time monitoring and control of the cutting die processing quality can be achieved, thereby improving production efficiency and ensuring product quality.

[0040] In an embodiment of the present application, the superpixel segmentation module 130 is used to perform superpixel segmentation on the surface state image of the knife mold in the molding quality monitoring server to obtain a sequence of superpixel blocks of the surface state of the knife mold. It should be understood that different areas in the surface state image of the knife mold have similar surface feature information, such as texture, structure, etc. Based on this, in order to focus more on the surface state of similar areas in the surface state image of the knife mold, in the technical solution of the present application, the surface state image of the knife mold is superpixel segmented in the molding quality monitoring server to obtain a sequence of superpixel blocks of the surface state of the knife mold. That is, the superpixel segmentation can divide the surface state image into multiple area blocks with similar features, and these area blocks are consistent in texture, brightness, etc., and can better retain the semantic information in the surface state image, which helps to better understand the content and structure between each superpixel block in subsequent processing.

[0041] In an embodiment of the present application, the superpixel block context semantic coding module 140 is used to perform superpixel block context semantic coding on the sequence of the die surface state superpixel blocks in the molding quality monitoring server to obtain a sequence of context die surface state superpixel block semantic coding feature vectors. Specifically, in an embodiment of the present application, the superpixel block context semantic coding module is used to: in the molding quality monitoring server, pass the sequence of the die surface state superpixel blocks through a superpixel block context semantic encoder including a superpixel block embedding coding module to obtain a sequence of context die surface state superpixel block semantic coding feature vectors. Accordingly, it is considered that the sequence of the die surface state superpixel blocks has context semantic information about each die surface state superpixel. Therefore, in order to capture and mine the contextual semantic feature information between each knife die surface state superpixel block, in the technical solution of the present application, in the molding quality monitoring server, the sequence of the knife die surface state superpixel blocks is passed through a superpixel block contextual semantic encoder including a superpixel block embedded coding module to obtain a sequence of contextual knife die surface state superpixel block semantic coding feature vectors. In this way, the information around each knife die surface state superpixel block can be used to better understand the semantic meaning represented by each knife die surface state superpixel block, so as to better capture the complex semantic information of the knife die surface state.

[0042] In an embodiment of the present application, the contextual semantic enhancement module 150 is used to input the sequence of contextual knife mold surface state super pixel block semantic encoding feature vectors into the contextual semantic enhancement module based on saliency-globality in the molding quality monitoring server to obtain a sequence of contextual enhanced knife mold surface state super pixel block semantic encoding feature vectors. It should be understood that considering that the sequence of contextual knife mold surface state super pixel block semantic encoding feature vectors contains the key saliency and overall global information of the semantic features of the global knife mold surface state super pixel block, these feature information have an important role and influence on the molding quality evaluation of the knife mold being processed and formed. Based on this, in order to improve and strengthen the ability to utilize these key saliency and overall features, thereby improving the representation ability of the semantic features of the entire contextual knife mold surface state super pixel block, in the technical solution of the present application, in the molding quality monitoring server, the sequence of contextual knife mold surface state super pixel block semantic encoding feature vectors is input into the contextual semantic enhancement module based on saliency-globality to obtain a sequence of contextual enhanced knife mold surface state super pixel block semantic encoding feature vectors. In detail, the contextual semantic enhancement module based on saliency-globality first extracts the maximum and average values ​​of the semantic encoding feature vectors of each contextual knife die surface state super pixel block in the sequence of the semantic encoding feature vectors of the contextual knife die surface state super pixel block to obtain salient and global features, so as to respectively highlight the most important and significant features of each super pixel block describing the knife die surface state and to globally understand the semantics of the entire knife die surface state. Then, the obtained salient and global features are respectively subjected to convolution activation and nonlinear processing to extract the semantic encoding features of the higher-level surface state super pixel block to provide richer semantic information. Then, the salient and global semantic features obtained after nonlinear processing are fused to make the pixel block semantic features more expressive and semantically rich. Finally, the fused salient-global fusion features are used as weights to weight the sequence of the contextual knife die surface state super pixel block semantic encoding feature vectors vector by vector, so that the model pays more attention to the important semantic information of the pixel block represented by the salient and global features, thereby improving the model's attention to the semantic features of the salient pixel blocks.

[0043] Specifically, in the embodiment of the present application, the context semantic enhancement module includes: a prominent feature calculation unit, which is used to extract the maximum value of the semantic coding feature vectors of each context knife die surface state super pixel block in the sequence of the context knife die surface state super pixel block semantic coding feature vector to obtain a prominent context knife die surface state super pixel block semantic coding feature vector; a global feature calculation unit, which is used to extract the average value of the semantic coding feature vectors of each context knife die surface state super pixel block in the sequence of the context knife die surface state super pixel block semantic coding feature vector to obtain a global context knife die surface state super pixel block semantic coding feature vector; a prominent feature one-dimensional convolution coding unit, which is used to convert the prominent context knife die surface state super pixel block semantic coding feature vector into a global context knife die surface state super pixel block semantic coding feature vector; The prominent context knife die surface state super pixel block convolution coding vector obtained by performing one-dimensional convolution coding on the semantic coding feature vector of the knife die surface state super pixel block below is input into the ReLU function to obtain the prominent context knife die surface state super pixel block activation coding feature vector; the prominent feature activation unit is used to perform matrix multiplication of the prominent point convolution feature vector obtained by performing point convolution coding on the prominent context knife die surface state super pixel block activation coding feature vector with the first weight matrix to obtain the prominent feature context knife die surface state super pixel block semantic association activation feature vector; the global feature one-dimensional convolution coding unit is used to perform one-dimensional convolution coding on the global context knife die surface state super pixel block semantic coding feature vector to obtain The global context knife mold surface state super pixel block convolution coding vector is input into the ReLU function to obtain the global context knife mold surface state super pixel block activation coding feature vector; the global feature activation unit is used to perform matrix multiplication of the global point convolution feature vector obtained after point convolution coding the global context knife mold surface state super pixel block activation coding feature vector with the second weight matrix to obtain the global feature context knife mold surface state super pixel block semantic association activation feature vector; the feature fusion unit is used to fuse the salient feature context knife mold surface state super pixel block semantic association activation feature vector and the global feature context knife mold surface state super pixel block semantic association activation feature vector to obtain the global-salient feature context After obtaining the context knife die surface state superpixel block semantic association feature vector, the global-salient feature context knife die surface state superpixel block semantic association feature vector is nonlinearly activated to obtain the global-salient feature context knife die surface state superpixel block semantic association weight feature vector; and a feature weighting unit is used to use the global-salient feature context knife die surface state superpixel block semantic association weight feature vector as a weight, perform positional dot multiplication on the sequence of context knife die surface state superpixel block semantic coding feature vectors and add the sequence of context knife die surface state superpixel block semantic coding feature vectors to obtain the sequence of context enhanced knife die surface state superpixel block semantic coding feature vectors.

[0044] More specifically, in an embodiment of the present application, the feature fusion unit is used to: perform feature fusion on the salient feature context knife mold surface state superpixel block semantic association activation feature vector and the global feature context knife mold surface state superpixel block semantic association activation feature vector to obtain the global-salient feature context knife mold surface state superpixel block semantic association feature vector; input the global-salient feature context knife mold surface state superpixel block semantic association feature vector into the tanh function to obtain a first activated global-salient feature context knife mold surface state superpixel block semantic association feature vector; input the global-salient feature context knife mold surface state superpixel block semantic association feature vector into the sigmoid function to obtain a second activated global-salient feature context knife mold surface state superpixel block semantic association feature vector; and perform position point multiplication on the first activated global-salient feature context knife mold surface state superpixel block semantic association feature vector and the second activated global-salient feature context knife mold surface state superpixel block semantic association feature vector to obtain the global-salient feature context knife mold surface state superpixel block semantic association weight feature vector.

[0045] In the embodiment of the present application, specifically, the contextual semantic enhancement module is used to: use the contextual semantic enhancement module based on saliency-globality to process the sequence of semantic encoding feature vectors of the contextual knife die surface state super pixel block with the following semantic enhancement formula to obtain the sequence of contextual enhanced knife die surface state super pixel block semantic encoding feature vectors; wherein, the semantic enhancement formula is: in, is a sequence of semantically encoded feature vectors of the contextual knife die surface state superpixel block, and are the maximum value and the average value of each eigenvector in the sequence of eigenvectors, and are respectively the semantic encoding feature vector of the salient context knife die surface state super pixel block and the semantic encoding feature vector of the global context knife die surface state super pixel block, is a one-dimensional convolutional code, yes function, is point convolutional coding, and are the first weight matrix and the second weight matrix respectively, and are respectively the semantically associated activation feature vector of the superpixel block of the salient feature context knife die surface state and the semantically associated activation feature vector of the superpixel block of the global feature context knife die surface state, is the function, yes function, It means point multiplication by position. is the global-salient feature context die surface state superpixel block semantic association weight feature vector, It is a sequence of semantically encoded feature vectors of superpixel blocks of the context-enhanced die surface state.

[0046] In an embodiment of the present application, the global fusion module 160 is used to input the sequence of context-enhanced knife die surface state superpixel block semantic coding feature vectors into the energy metric attention fusion network based on sequence annotation in the molding quality monitoring server to obtain the knife die surface state global fusion feature vector as the knife die surface state global fusion feature. Accordingly, considering that different context-enhanced knife die surface state superpixel block semantic coding features in the sequence of context-enhanced knife die surface state superpixel block semantic coding feature vectors have different importance and contribution to the entire image. Therefore, in order to further capture the context knife die superpixel block correlation characteristics and significance information represented by each context-enhanced knife die surface state superpixel block semantic coding feature vector, in the technical solution of the present application, in the molding quality monitoring server, the sequence of context-enhanced knife die surface state superpixel block semantic coding feature vectors is input into the energy metric attention fusion network based on sequence annotation to obtain the knife die surface state global fusion feature vector. In detail, the energy measurement attention fusion network based on sequence labeling labels each context-enhanced knife die surface state superpixel block semantic coding feature vector in the sequence of the context-enhanced knife die surface state superpixel block semantic coding feature vector with a serial number to emphasize the semantic importance and meaning of the superpixel block of each vector in the sequence. Then, the variance and mean of each context-enhanced knife die surface state superpixel block semantic coding vector are calculated, and the energy function value of each context-enhanced knife die surface state superpixel block semantic coding vector is calculated based on the variance, mean and serial number, so as to perform a corresponding importance evaluation on each context-enhanced knife die surface state superpixel block semantic coding vector. Finally, based on each energy function value, weighted fusion is performed on each context-enhanced knife die surface state superpixel block semantic coding image vector to more comprehensively and emphatically represent the semantic information of the entire knife die surface state.

[0047] Specifically, in an embodiment of the present application, the global fusion module includes: a vector serial number labeling unit, which is used to label the serial number of each context-enhanced knife die surface state super pixel block semantic coding feature vector in the sequence of the context-enhanced knife die surface state super pixel block semantic coding feature vector to obtain a sequence of context-enhanced knife die surface state super pixel block semantic labeling serial numbers; a vector energy value calculation unit, which is used to calculate the energy function value of each context-enhanced knife die surface state super pixel block semantic coding feature vector in the sequence of the context-enhanced knife die surface state super pixel block semantic coding feature vector to obtain a sequence of context-enhanced knife die surface state super pixel block semantic energy function values, wherein the energy function value of each context-enhanced knife die surface state super pixel block semantic coding feature vector is related to the mean and variance of each context-enhanced knife die surface state super pixel block semantic coding feature vector and the semantic labeling serial number of each context-enhanced knife die surface state super pixel block; A function value activation unit is used to input the inverse of each context-enhanced knife die surface state superpixel block semantic energy function value in the sequence of context-enhanced knife die surface state superpixel block semantic energy function values ​​into a sigmoid function to obtain a sequence of context-enhanced knife die surface state superpixel block semantic activation energy function values; a feature position point multiplication unit is used to use each context-enhanced knife die surface state superpixel block semantic activation energy function value in the sequence of context-enhanced knife die surface state superpixel block semantic activation energy function values ​​as a weight, and perform position point multiplication on each context-enhanced knife die surface state superpixel block semantic coding feature vector in the sequence of context-enhanced knife die surface state superpixel block semantic coding feature vectors to obtain a sequence of context-salient knife die surface state superpixel block semantic coding feature vectors; and a fusion unit is used to fuse the sequence of context-salient knife die surface state superpixel block semantic coding feature vectors to obtain the knife die surface state global fusion feature vector.

[0048] More specifically, in an embodiment of the present application, the vector energy value calculation unit is used to: calculate the mean of the context-enhanced knife die surface state superpixel block semantic coding feature vector to obtain the context-enhanced knife die surface state superpixel block semantic mean; calculate the variance of the context-enhanced knife die surface state superpixel block semantic coding feature vector to obtain the context-enhanced knife die surface state superpixel block semantic variance; multiply the value obtained by adding the context-enhanced knife die surface state superpixel block semantic variance and the regularization term parameter by four to obtain the context-enhanced knife die surface state superpixel block semantic first energy function value; calculate the context-enhanced knife die surface state superpixel block semantic labeling serial number corresponding to the context-enhanced knife die surface state superpixel block semantic coding feature vector The square of the difference between the context-enhanced knife die surface state super pixel block semantic mean is used to obtain the context-enhanced knife die surface state super pixel block semantic offset value; the modulated context-enhanced knife die surface state super pixel block semantic variance obtained by multiplying the context-enhanced knife die surface state super pixel block semantic variance by two is added to the value obtained by multiplying the regularization term parameter by two and the context-enhanced knife die surface state super pixel block semantic offset value to obtain the context-enhanced knife die surface state super pixel block semantic second energy function value; and the context-enhanced knife die surface state super pixel block semantic first energy function value is divided by the context-enhanced knife die surface state super pixel block semantic second energy function value to obtain the context-enhanced knife die surface state super pixel block semantic energy function value.

[0049] In the embodiment of the present application, specifically, the vector energy value calculation unit is used to: process the semantic coding feature vector of the context-enhanced knife die surface state super pixel block according to the following energy function calculation formula to obtain the semantic energy function value of the context-enhanced knife die surface state super pixel block; wherein the energy function calculation formula is:

[0050]

[0051]

[0052] in, The context-enhanced die surface state superpixel block semantic encoding feature vector is The eigenvalues ​​at the positions, is the number of eigenvalues ​​in the semantic encoding feature vector of the context-enhanced die surface state superpixel block, is the semantic mean of the superpixel block of the context-enhanced die surface state, is the context-enhanced superpixel semantic variance of the surface state of the die, is the regularization parameter, is the semantic annotation sequence number of the context-enhanced die surface state superpixel block, It is the semantic energy function value of the superpixel block of the context-enhanced knife die surface state.

[0053] In an embodiment of the present application, the quality inspection result generation module 170 is used to obtain a quality inspection result based on the global fusion feature of the surface state of the knife die in the molding quality monitoring server, and the quality inspection result is used to indicate whether the molding quality of the processed knife die meets the predetermined standard. Specifically, in an embodiment of the present application, the quality inspection result generation module is used to: input the global fusion feature vector of the surface state of the knife die into a quality inspection result generator based on a classifier to obtain a quality inspection result. That is, the global fusion feature of the surface state of the knife die obtained by attention fusion of the sequence of semantic encoding feature vectors of superpixel blocks of the context-enhanced surface state of the knife die is used for classification processing, so as to automatically determine whether the molding quality of the processed knife die meets the predetermined standard. In this way, the molding quality monitoring server can be used to quickly analyze the surface state of the knife die, which greatly improves the detection speed and the accuracy of quality control compared to manual inspection.

[0054] It is worth mentioning that those skilled in the art should be aware that before applying a deep neural network model for inference, the deep neural network model must be trained so that the deep neural network can implement specific functional capabilities.

[0055] Specifically, in the technical solution of the present application, the knife die processing equipment with high profiling sharpness and integrated waste removal also includes a training module for training the superpixel block context semantic encoder including the superpixel block embedding coding module, the saliency-global based context semantic enhancement module, the sequence labeling based energy measurement attention fusion network and the classifier based quality inspection result generator.

[0056] Figure 3 1 is a block diagram of a training module in a die-cutting processing device with high profiling sharpness and integrated waste discharge according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 3As shown, the training module 200 includes: a training data acquisition unit 210, which is used to acquire training data, wherein the training data includes a surface state image of a training die, and a real quality inspection result, wherein the real quality inspection result is a real value of whether the molding quality of the processed die meets a predetermined standard; a training image transmission unit 220, which is used to transmit the surface state image of the training die to a molding quality monitoring server of the die processing equipment; a training image segmentation unit 230, which is used to perform superpixel segmentation on the surface state image of the training die at the molding quality monitoring server to obtain a sequence of superpixel blocks of the surface state of the training die; a training superpixel block semantic encoding unit 240, which is used to pass the sequence of superpixel blocks of the surface state of the training die through the superpixel block context semantic encoder including the superpixel block embedding encoding module at the molding quality monitoring server to obtain a sequence of semantic encoding feature vectors of the superpixel blocks of the surface state of the training die; and a training semantic enhancement unit 250, which is used to input the sequence of semantic encoding feature vectors of the superpixel blocks of the surface state of the training die into the saliency-based superpixel block semantic encoding feature vector at the molding quality monitoring server. - a global context semantic enhancement module to obtain a sequence of training context enhanced superpixel block semantic encoding feature vectors of the surface state of the die; a training feature fusion unit 260, used to input the sequence of training context enhanced superpixel block semantic encoding feature vectors of the surface state of the die into the energy metric attention fusion network based on sequence annotation in the molding quality monitoring server to obtain a global fusion feature vector of the surface state of the die; a training quality inspection result generation unit 270, used to input the global fusion feature vector of the surface state of the die into the quality inspection result generator based on the classifier to obtain a training quality inspection result; a classification loss unit 280, used to calculate the cross entropy loss function value between the training quality inspection result and the real quality inspection result to obtain a classification loss function value; and a model training unit 290, used to train the superpixel block context semantic encoder including the superpixel block embedding coding module, the saliency-global context semantic enhancement module, the energy metric attention fusion network based on sequence annotation and the quality inspection result generator based on the classifier based on the classification loss function value and through gradient descent back propagation.

[0057] It should be understood that, here, for each training context knife die surface state superpixel block semantic encoding feature vector in the sequence of the training context knife die surface state superpixel block semantic encoding feature vectors, the image semantic features of the surface state image of the knife die in the local image semantic space domain are respectively expressed. After the image semantic space domain inter-domain saliency-global context semantic enhancement and energy measurement attention fusion in the global image semantic space domain, due to the difference in image semantic feature distribution in the local image semantic space domain, the saliency-global enhancement weight distribution and energy measurement attention weight difference caused by the difference in image semantic feature distribution in the local image semantic space domain, the training knife die surface state global fusion feature vector will also have the problem of insufficient aggregation of fused image semantic feature distribution, thereby affecting its classification convergence efficiency through the classifier-based quality inspection result generator, that is, affecting the efficiency of classification training and the accuracy of classification results.

[0058] Preferably, the global fusion feature vector of the training knife die surface state is input into the classifier-based quality inspection result generator to obtain the training quality inspection result, including: multiplying each eigenvalue of the global fusion feature vector of the training knife die surface state by the length of the global fusion feature vector of the training knife die surface state, subtracting the first norm of the global fusion feature vector of the training knife die surface state, and calculating the square root of the absolute value of the subtraction result to obtain the first training knife die surface state global fusion intermediate vector; multiplying each eigenvalue of the global fusion feature vector of the training knife die surface state by the square root of the length of the global fusion feature vector of the training knife die surface state, subtracting the second norm of the global fusion feature vector of the training knife die surface state, and calculating the square root of the absolute value of the subtraction result to obtain to the second training knife die surface state global fusion intermediate vector; respectively calculate the base 2 logarithm of each eigenvalue of the first training knife die surface state global fusion intermediate vector and the second training knife die surface state global fusion intermediate vector to obtain the first training knife die surface state global fusion intermediate information vector and the second training knife die surface state global fusion intermediate information vector; calculate the weighted sum of the first training knife die surface state global fusion intermediate information vector and the second training knife die surface state global fusion intermediate information vector with the weighting coefficient as a hyperparameter to obtain the corrected training knife die surface state global fusion feature vector; and, input the corrected training knife die surface state global fusion feature vector into the classifier-based quality inspection result generator to obtain the training quality inspection result.

[0059] Specifically, in this preferred embodiment, the global fusion feature vector of the surface state of the training die is input into the classifier-based quality inspection result generator to obtain the training quality inspection result, including:

[0060] The global fusion feature vector of the surface state of the training die is corrected to obtain the corrected global fusion feature vector of the surface state of the training die. The process formula is as follows:

[0061]

[0062] in, is the global fusion feature vector of the surface state of the training die, is the length of the global fusion feature vector of the surface state of the training die, is the one-norm of the eigenvector, It is the point multiplication by position. It is subtracted by position point. is the bi-norm of the eigenvector, is the hyperparameter, is added by position, It means calculating the square root of the absolute value of the eigenvalues ​​at each position in the eigenvector. represents the logarithmic function value with base 2, is the global fusion feature vector of the corrected training die surface state;

[0063] The corrected global fusion feature vector of the surface state of the training die is input into the classifier-based quality inspection result generator to obtain the training quality inspection result.

[0064] Based on this, the global fusion feature vector of the training die surface state is used as a feature set, wherein the change semantics represented by the feature value of the global fusion feature vector of the training die surface state is used as a unit. In order to dynamically aggregate the semantic set composed of different change semantics of the global fusion feature vector of the training die surface state without ignoring the individual semantic change information, the individual features of the global fusion feature vector of the training die surface state and its aggregated scale are expressed as the feature full amplitude and half amplitude, and the low-rank negative correlation of the overall semantics of the feature set of the global fusion feature vector of the training die surface state is used as the phase and scaled to dynamically adjust the change relationship of the semantic content of the global fusion feature vector of the training die surface state, thereby improving the overall image semantic information expression aggregation of the feature set of the global fusion feature vector of the training die surface state, and improving the classification convergence efficiency of the global fusion feature vector of the training die surface state through the quality inspection result generator based on the classifier, that is, improving the efficiency of classification training and the accuracy of quality inspection results. In this way, the molding quality monitoring server can be used to quickly analyze the surface state of the die, which greatly improves the detection speed and quality control accuracy compared to manual inspection.

[0065] In summary, the knife die processing equipment 100 with high profiling sharpness and integrated waste discharge based on the embodiment of the present application is explained, which transmits the surface state image of the processed knife die collected by the industrial camera deployed in the knife die processing equipment to the molding quality monitoring server of the knife die processing equipment, and uses AI-based image recognition and processing technology in the molding quality monitoring server to perform super-pixel segmentation and encoding on the knife die surface state image, so as to automatically judge whether the molding quality of the processed knife die meets the predetermined standard according to the fusion semantic representation features of each super-pixel block in the whole area. In this way, the molding quality monitoring server can be used to quickly analyze the surface state of the knife die, which greatly improves the detection speed and quality control accuracy compared with manual inspection.

[0066] As described above, the knife die processing device 100 with high profiling sharpness and integrated waste removal according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a knife die processing algorithm with high profiling sharpness and integrated waste removal. In a possible implementation, the knife die processing device 100 with high profiling sharpness and integrated waste removal according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the knife die processing device 100 with high profiling sharpness and integrated waste removal can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the knife die processing device 100 with high profiling sharpness and integrated waste removal can also be one of the many hardware modules of the wireless terminal.

[0067] Alternatively, in another example, the knife mold processing equipment with high profiling sharpness and integrated waste discharge 100 and the wireless terminal can also be discrete devices, and the knife mold processing equipment with high profiling sharpness and integrated waste discharge 100 can be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0068] Figure 4 Flow chart of a cutting die processing method with high profiling sharpness and integrated waste removal according to an embodiment of the present application. Figure 4As shown, according to the embodiment of the present application, the knife die processing method with high profiling sharpness and integrated waste discharge includes: S110, collecting the surface state image of the processed knife die by an industrial camera deployed on the knife die processing equipment; S120, transmitting the surface state image of the knife die to the molding quality monitoring server of the knife die processing equipment; S130, in the molding quality monitoring server, performing super-pixel segmentation on the surface state image of the knife die to obtain a sequence of knife die surface state super-pixel blocks; S140, in the molding quality monitoring server, performing super-pixel block context semantic coding on the sequence of knife die surface state super-pixel blocks to obtain a sequence of context knife die surface state super-pixel block semantic coding feature vectors; S150, in the molding quality monitoring server In the server, the sequence of contextual knife die surface state superpixel block semantic coding feature vectors is input into the contextual semantic enhancement module based on significance-globality to obtain the sequence of contextual enhanced knife die surface state superpixel block semantic coding feature vectors; S160, in the molding quality monitoring server, the sequence of contextual enhanced knife die surface state superpixel block semantic coding feature vectors is input into the energy measurement attention fusion network based on sequence annotation to obtain the knife die surface state global fusion feature vector as the knife die surface state global fusion feature; and, S170, in the molding quality monitoring server, the quality inspection result is obtained based on the knife die surface state global fusion feature, and the quality inspection result is used to indicate whether the molding quality of the processed knife die meets the predetermined standard.

[0069] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned high-profile sharpness and integrated waste discharge die processing method have been referenced above. Figures 1 to 3 The description of the high profiling sharpness and integrated waste discharge die-cutting processing equipment has been introduced in detail, and therefore, its repeated description will be omitted.

[0070] The above descriptions of various implementations of the present disclosure are exemplary and not exhaustive. The disclosure is not limited to the disclosed implementations, and many modifications and variations are obvious to those of ordinary skill in the art without departing from the scope and spirit of the described implementations. The terms used herein are selected to best explain the principles of the implementations, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the various practical methods disclosed herein.

Claims

1. A die-cutting processing equipment with high profiling sharpness and integrated waste discharge, characterized in that: include: The die cutting surface state image acquisition module is used to acquire the surface state image of the die cutting mold through the industrial camera deployed on the die cutting processing equipment; An image transmission module, used for transmitting the surface state image of the cutting die to a molding quality monitoring server of the cutting die processing equipment; A superpixel segmentation module is used to perform superpixel segmentation on the surface state image of the knife die on the molding quality monitoring server to obtain a sequence of superpixel blocks of the surface state of the knife die; A superpixel block context semantic coding module is used to perform superpixel block context semantic coding on the sequence of the die surface state superpixel blocks on the molding quality monitoring server to obtain a sequence of context die surface state superpixel block semantic coding feature vectors; A context semantic enhancement module is used to input the sequence of context die surface state super pixel block semantic encoding feature vectors into the context semantic enhancement module based on saliency-globality in the molding quality monitoring server to obtain a sequence of context enhanced die surface state super pixel block semantic encoding feature vectors; A global fusion module is used for inputting the sequence of context-enhanced superpixel semantic encoding feature vectors of the surface state of the die into the energy measurement attention fusion network based on sequence annotation on the molding quality monitoring server to obtain a global fusion feature vector of the surface state of the die as a global fusion feature of the surface state of the die; A quality inspection result generating module is used to obtain a quality inspection result based on the global fusion feature of the surface state of the die cutter on the molding quality monitoring server, wherein the quality inspection result is used to indicate whether the molding quality of the die cutter used for molding meets the predetermined standard; Wherein, the context semantics enhancement module includes: A salient feature calculation unit, used for extracting the maximum value of each contextual knife die surface state super pixel block semantic coding feature vector in the sequence of contextual knife die surface state super pixel block semantic coding feature vector to obtain a salient contextual knife die surface state super pixel block semantic coding feature vector; A global feature calculation unit, used for extracting the average value of the semantic coding feature vectors of each contextual knife die surface state super pixel block in the sequence of the contextual knife die surface state super pixel block semantic coding feature vector to obtain a global contextual knife die surface state super pixel block semantic coding feature vector; A salient feature one-dimensional convolutional coding unit is used to input the salient context knife die surface state super pixel block convolutional coding vector obtained after one-dimensional convolutional coding of the salient context knife die surface state super pixel block semantic coding feature vector into a ReLU function to obtain a salient context knife die surface state super pixel block activation coding feature vector; A salient feature activation unit is used to perform matrix multiplication of a salient point convolution feature vector obtained by performing point convolution encoding on the salient context die surface state super pixel block activation encoding feature vector and a first weight matrix to obtain a salient feature context die surface state super pixel block semantic association activation feature vector; A global feature one-dimensional convolutional coding unit, used for inputting the global context knife die surface state super pixel block convolutional coding vector obtained by performing one-dimensional convolution coding on the global context knife die surface state super pixel block semantic coding feature vector into a ReLU function to obtain a global context knife die surface state super pixel block activation coding feature vector; A global feature activation unit, used for performing matrix multiplication of a global point convolution feature vector obtained by performing point convolution encoding on the global context die surface state super pixel block activation encoding feature vector and a second weight matrix to obtain a global feature context die surface state super pixel block semantic association activation feature vector; A feature fusion unit is used for fusing the salient feature context die surface state super pixel block semantic association activation feature vector and the global feature context die surface state super pixel block semantic association activation feature vector to obtain a global-salient feature context die surface state super pixel block semantic association feature vector, and then performing nonlinear activation on the global-salient feature context die surface state super pixel block semantic association feature vector to obtain a global-salient feature context die surface state super pixel block semantic association weight feature vector; A feature weighting unit is used to use the global-salient feature context knife die surface state super pixel block semantic association weight feature vector as a weight, perform positional dot multiplication on the sequence of context knife die surface state super pixel block semantic coding feature vectors, and add the sequence of context knife die surface state super pixel block semantic coding feature vectors to obtain the sequence of context enhanced knife die surface state super pixel block semantic coding feature vectors.

2. The die-cutting processing equipment with high profiling sharpness and integrated waste discharge according to claim 1 is characterized in that: The superpixel block context semantic encoding module is used to: in the molding quality monitoring server, pass the sequence of the superpixel blocks of the die surface state through a superpixel block context semantic encoder including a superpixel block embedding encoding module to obtain a sequence of context die surface state superpixel block semantic encoding feature vectors.

3. The die-cutting processing equipment with high profiling sharpness and integrated waste discharge according to claim 2 is characterized in that: The feature fusion unit is used to: Performing feature fusion on the semantic association activation feature vector of the salient feature context knife die surface state super pixel block and the semantic association activation feature vector of the global feature context knife die surface state super pixel block to obtain the global-salient feature context knife die surface state super pixel block semantic association feature vector; Inputting the global-salient feature context knife die surface state super pixel block semantic association feature vector into a tanh function to obtain a first activated global-salient feature context knife die surface state super pixel block semantic association feature vector; Inputting the global-salient feature context knife die surface state super pixel block semantic association feature vector into a Sigmoid function to obtain a second activated global-salient feature context knife die surface state super pixel block semantic association feature vector; The first activated global-salient feature context knife mold surface state super pixel block semantic association feature vector and the second activated global-salient feature context knife mold surface state super pixel block semantic association feature vector are point-wise multiplied to obtain the global-salient feature context knife mold surface state super pixel block semantic association weight feature vector.

4. The die-cutting processing equipment with high profiling sharpness and integrated waste discharge according to claim 3 is characterized in that: The global fusion module comprises: A vector serial number labeling unit is used to label each context-enhanced knife die surface state super pixel block semantic coding feature vector in the sequence of context-enhanced knife die surface state super pixel block semantic coding feature vectors with a serial number to obtain a sequence of context-enhanced knife die surface state super pixel block semantic labeling serial numbers; A vector energy value calculation unit, used for calculating the energy function value of each context-enhanced knife die surface state superpixel block semantic coding feature vector in the sequence of context-enhanced knife die surface state superpixel block semantic coding feature vectors to obtain a sequence of context-enhanced knife die surface state superpixel block semantic energy function values, wherein the energy function value of each context-enhanced knife die surface state superpixel block semantic coding feature vector is related to the mean and variance of each context-enhanced knife die surface state superpixel block semantic coding feature vector and the semantic labeling sequence number of each context-enhanced knife die surface state superpixel block; An energy function value activation unit is used to input the reciprocal of each context-enhanced knife die surface state super pixel block semantic energy function value in the sequence of context-enhanced knife die surface state super pixel block semantic energy function values ​​into a sigmoid function to obtain a sequence of context-enhanced knife die surface state super pixel block semantic activation energy function values; A feature point multiplication unit is used for performing point multiplication by position on each context-enhanced die surface state superpixel block semantic coding feature vector in the sequence of context-enhanced die surface state superpixel block semantic coding feature vectors, using each context-enhanced die surface state superpixel block semantic activation energy function value in the sequence of context-enhanced die surface state superpixel block semantic activation energy function value as a weight, so as to obtain a sequence of context-salient die surface state superpixel block semantic coding feature vectors; A fusion unit is used to fuse the sequence of context-significant superpixel block semantic encoding feature vectors of the surface state of the die to obtain the global fusion feature vector of the surface state of the die.

5. The die-cutting processing equipment with high profiling sharpness and integrated waste discharge according to claim 4 is characterized in that: The vector energy value calculation unit is used to: Calculating the mean of the semantic encoding feature vector of the context-enhanced knife die surface state super pixel block to obtain the context-enhanced knife die surface state super pixel block semantic mean; Calculating the variance of the semantic encoding feature vector of the context-enhanced knife die surface state super pixel block to obtain the context-enhanced knife die surface state super pixel block semantic variance; The value obtained by adding the context-enhanced super pixel block semantic variance of the surface state of the knife die and the regularization term parameter is multiplied by four to obtain the context-enhanced super pixel block semantic first energy function value of the surface state of the knife die; Calculate the square of the difference between the context-enhanced knife die surface state super pixel block semantic labeling sequence number corresponding to the context-enhanced knife die surface state super pixel block semantic encoding feature vector and the context-enhanced knife die surface state super pixel block semantic mean to obtain the context-enhanced knife die surface state super pixel block semantic offset value; The context-enhanced knife die surface state super pixel block semantic variance obtained by multiplying the context-enhanced knife die surface state super pixel block semantic variance by two is added to the value obtained by multiplying the regularization term parameter by two and the context-enhanced knife die surface state super pixel block semantic offset value to obtain the context-enhanced knife die surface state super pixel block semantic second energy function value; The context-enhanced knife die surface state super pixel block semantic first energy function value is divided by the context-enhanced knife die surface state super pixel block semantic second energy function value to obtain the context-enhanced knife die surface state super pixel block semantic energy function value.

6. The die-cutting processing equipment with high profiling sharpness and integrated waste discharge according to claim 5 is characterized in that: The quality inspection result generating module is used to: input the global fusion feature vector of the surface state of the cutting die into the quality inspection result generator based on the classifier to obtain the quality inspection result.

7. The die-cutting processing equipment with high profiling sharpness and integrated waste discharge according to claim 6 is characterized in that: It also includes a training module for training the superpixel block context semantic encoder including the superpixel block embedding coding module, the saliency-global based context semantic enhancement module, the sequence labeling based energy measurement attention fusion network and the classifier based quality inspection result generator.

8. The die-cutting processing equipment with high profiling sharpness and integrated waste discharge according to claim 7 is characterized in that: The training module comprises: A training data acquisition unit, used to acquire training data, wherein the training data includes a surface state image of a training cutter die and a real quality inspection result, wherein the real quality inspection result is a real value of whether the molding quality of the processed cutter die meets a predetermined standard; A training image transmission unit, used for transmitting the surface state image of the training cutting die to a molding quality monitoring server of the cutting die processing equipment; A training image segmentation unit is used to perform super-pixel segmentation on the surface state image of the training die on the molding quality monitoring server to obtain a sequence of super-pixel blocks of the surface state of the training die; A training superpixel block semantic encoding unit is used to pass the sequence of the training die surface state superpixel blocks through the superpixel block context semantic encoder including the superpixel block embedding encoding module on the molding quality monitoring server to obtain a sequence of training context die surface state superpixel block semantic encoding feature vectors; A training semantic enhancement unit is used to input the sequence of semantic encoding feature vectors of the super pixel block of the training context die surface state into the context semantic enhancement module based on saliency-globality in the molding quality monitoring server to obtain a sequence of semantic encoding feature vectors of the super pixel block of the training context enhanced die surface state; A training feature fusion unit is used to input the sequence of the training context enhanced superpixel block semantic encoding feature vectors of the surface state of the die into the energy measurement attention fusion network based on sequence annotation to obtain a global fusion feature vector of the surface state of the training die in the molding quality monitoring server; A training quality inspection result generating unit, used for inputting the global fusion feature vector of the surface state of the training die into the classifier-based quality inspection result generator to obtain a training quality inspection result; A classification loss unit, used to calculate a cross entropy loss function value between the training quality inspection result and the real quality inspection result to obtain a classification loss function value; A model training unit is used to train the superpixel block context semantic encoder including the superpixel block embedding coding module, the saliency-global based context semantic enhancement module, the sequence labeling based energy measurement attention fusion network and the classifier based quality inspection result generator based on the classification loss function value and through gradient descent back propagation.

9. A method for processing a die cutter with high profiling sharpness and integrated waste removal, using the die cutter processing equipment with high profiling sharpness and integrated waste removal as claimed in claim 1, characterized in that: include: The surface state image of the processed die is collected by an industrial camera deployed in the die processing equipment; Transmitting the surface state image of the cutting die to a molding quality monitoring server of the cutting die processing equipment; On the molding quality monitoring server, super-pixel segmentation is performed on the surface state image of the knife die to obtain a sequence of super-pixel blocks of the surface state of the knife die; On the molding quality monitoring server, the sequence of super pixel blocks of the die surface state is subjected to super pixel block context semantic coding to obtain a sequence of context die surface state super pixel block semantic coding feature vectors; In the molding quality monitoring server, the sequence of contextual die surface state super pixel block semantic coding feature vectors is input into a contextual semantic enhancement module based on saliency-globality to obtain a sequence of contextual enhanced die surface state super pixel block semantic coding feature vectors; In the molding quality monitoring server, the sequence of context-enhanced superpixel semantic encoding feature vectors of the die surface state is input into an energy-metric attention fusion network based on sequence annotation to obtain a die surface state global fusion feature vector as a die surface state global fusion feature; In the molding quality monitoring server, a quality inspection result is obtained based on the global fusion feature of the surface state of the cutting die, and the quality inspection result is used to indicate whether the molding quality of the processed cutting die meets the predetermined standard.

Images