Method, device, equipment and medium for performing commodity material fission on commodity image

Abstract

The invention provides a commodity material fission method and device based on a commodity image, equipment and a medium, and the method comprises the steps: carrying out the judgment of an obtained image, and dividing the image into a pure commodity image or a figure commodity image; obtaining a corresponding commodity main body mask graph; if the commodity graph is a pure commodity graph, calling a canny model for extracting the edge of the commodity main body of the commodity main body mask graph through a graph-to-graph algorithm; the graph-to-graph algorithm calls an ipadatter algorithm to obtain element features in the pure commodity graph; redrawing the mask to obtain a generated graph; if the graph is a commodity graph with characters, calling a canny model by a graph-to-graph algorithm to perform line extraction on a commodity main body and characters in the commodity graph with the characters; the graph generation algorithm is used for calling an ipadatter algorithm; the tile algorithm is called by the graph-to-graph algorithm; and redrawing the mask to obtain a generated graph, so that a high-quality material library can be constructed for training a commodity graph Lora model.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and particularly relates to a method, device, equipment and medium for splitting and multiplying commodity materials from commodity images. Background Art

[0002] In the e-commerce field, the training of the Lora model has extremely large requirements for commodity image data. The commodity image data needs to cover various commodity main bodies and diverse backgrounds so that high-precision training of the Lora model can be carried out. However, the actually collected data is not necessarily usable, so these data need to be screened, which is time-consuming and laborious and increases the cost of enterprises;

[0003] At the same time, in the process of shooting and designing commodity pictures, how to produce different commodity pictures requires designers or artists to think. They need a lot of commodity picture materials as references for their inspiration support. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method, device, equipment and medium for splitting and multiplying commodity materials from commodity images, which not only provides designers with extensive design inspiration and reference materials, but also constructs a high-quality material library for training the commodity picture Lora model.

[0005] In the first aspect, the present invention provides a method for splitting and multiplying commodity materials from commodity images, including the following steps:

[0006] Step 1, judge the obtained picture, and divide it into a pure commodity picture or a commodity picture with a person;

[0007] Step 2, obtain the corresponding commodity main body mask picture through the picture;

[0008] Step 3, if it is a pure commodity picture, the image-to-image algorithm calls the canny model to extract the edge of the commodity main body in the commodity main body mask picture; the image-to-image algorithm calls the ipadapter algorithm to obtain the element features in the pure commodity picture; the commodity main body mask picture and the pure commodity picture are input into the image-to-image algorithm for mask redrawing to obtain a generated picture;

[0009] If it is a commodity picture with a person, the image-to-image algorithm calls the canny model to extract the lines of the commodity main body and the person in the commodity picture with a person; the image-to-image algorithm calls the ipadapter algorithm; the image-to-image algorithm calls the tile algorithm; the commodity main body mask picture and the pure commodity picture are input into the image-to-image algorithm for mask redrawing to obtain a generated picture.

[0010] In the second aspect, the present invention provides a device for splitting and multiplying commodity materials from commodity images, including:

[0011] The image classification module judges the acquired images and classifies them into pure product images or product images with people.

[0012] The product main body mask module obtains the corresponding product main body mask image through the said image.

[0013] The fission generation module, if it is a pure product image, the image-to-image algorithm calls the canny model to extract the edges of the product main body in the product main body mask image; the image-to-image algorithm calls the ipadapter algorithm to obtain the element features in the pure product image; the product main body mask image and the pure product image are passed into the image-to-image algorithm for mask redrawing to obtain the generated image.

[0014] If it is a product image with people, the image-to-image algorithm calls the canny model to extract the lines of the product main body and people in the product image with people; the image-to-image algorithm calls the ipadapter algorithm; the image-to-image algorithm calls the tile algorithm; the product main body mask image and the pure product image are passed into the image-to-image algorithm for mask redrawing to obtain the generated image.

[0015] In a third aspect, the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the method described in the first aspect is implemented.

[0016] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method described in the first aspect is implemented.

[0017] One or more technical solutions provided by the present invention have at least the following technical effects or advantages:

[0018] The present invention can quickly generate product image backgrounds in various styles, greatly improving the efficiency of design work, while ensuring the consistency between images and the natural fusion of the background and the product. This not only provides designers with a wide range of design inspirations and reference materials, but also can build a high-quality material library for training the product image Lora model.

[0019] The above description is only an overview of the technical solution of the present invention. In order to be able to understand the technical means of the present invention more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention are hereinafter specifically exemplified. Description of the Drawings

[0020] The present invention will be further described below with reference to the accompanying drawings in conjunction with the embodiments.

[0021] Figure 1It is the flowchart of the method in Embodiment 1 of the present invention;

[0022] Figure 2 It is the structural schematic diagram of the device in Embodiment 2 of the present invention. Detailed implementation manners

[0023] The technical solutions in the embodiments of the present application generally have the following ideas:

[0024] Openpose: OpenPose is an open-source project for real-time multi-person pose estimation, which can detect each key point of the human body in images or videos. It uses deep learning technology to identify the human body contour and 25 main joints, and is widely used in motion capture, human limb detection, etc. OpenPose is known for its high efficiency, accuracy, and robustness, and can handle multi-person pose recognition tasks in complex environments.

[0025] Ipadapter: IP-Adapter is an image prompt adapter, aiming to improve the image generation ability of pre-trained text-to-image diffusion models. It enables the pre-trained text-to-image diffusion models to more effectively process image prompts through a decoupled cross-attention mechanism.

[0026] tile model: The tile model in ControlNet is a model used to improve the image quality details in Stable Diffusion. It enhances details and repairs images by processing images in blocks.

[0027] Canny: The Canny algorithm is a popular image processing technology mainly used for edge detection, that is, identifying the edges in images, and can extract clear and accurate edge information from images. Here, it is mainly combined with img2img image-to-image generation to limit the edges of the commodity main body, ensuring the consistency of the edge information of the commodity main body after background replacement and the edges of the commodity image to be replaced.

[0028] The algorithm is mainly implemented as follows:

[0029] (1) Input the commodity image, call the openpose model, establish a logical judgment based on the key points detected by the openpose model for the image data, and judge whether the image contains a person or a person's limb.

[0030] (2) Establish different image processing logics, and divide the commodity image into a pure commodity image and a commodity image with a person.

[0031] (3) Combine the incoming main mask of the product image. If it is a pure product image, call the canny model to extract the edges of the main body of the mask, set the weight of canny to 0.85, call the ipadapter algorithm to obtain the element features in the product image, set the weight of ipadapter to 0.7, and finally pass it into the image-to-image mask redrawing, and set the redrawing amplitude of the mask redrawing to 0.75.

[0032] (4) If it is a product image with a person, call the canny model to extract the lines of the original product image, set the canny weight to 0.5, call the ipadapter algorithm, set its weight to 0.9, call the tile algorithm, set its weight to 0.6, and finally pass it into the image-to-image mask redrawing, and set the redrawing amplitude of the mask redrawing to 0.6.

[0033] (5) Run image-to-image to obtain the final image.

[0034] This final image can be used as a reference for design or for the training of existing Lora models.

[0035] Example 1

[0036] As Figure 1 shown, this example provides a method for product material fission of product images, including the following steps:

[0037] Step 1: Judge the obtained picture and classify it into a pure product image or a product image with a person;

[0038] Step 2: Obtain the corresponding main mask image of the product through the picture;

[0039] Step 3: If it is a pure product image, the image-to-image algorithm calls the canny model to extract the edges of the product main body in the product main mask image; the image-to-image algorithm calls the ipadapter algorithm to obtain the element features in the pure product image; pass the product main mask image and the pure product image into the image-to-image algorithm for mask redrawing to obtain the generated image;

[0040] If it is a product image with a person, the image-to-image algorithm calls the canny model to extract the lines of the product main body and the person in the product image with a person; the image-to-image algorithm calls the ipadapter algorithm; the image-to-image algorithm calls the tile algorithm; pass the product main mask image and the pure product image into the image-to-image algorithm for mask redrawing to obtain the generated image.

[0041] In this embodiment, preferably, step 1 is specifically as follows: Use the OpenPose model to judge the acquired picture. The OpenPose model performs joint point detection on the picture. If the picture contains a person or a person's limb, it is a picture of a commodity with a person; otherwise, it is a pure commodity picture.

[0042] In this embodiment, preferably, step 2 is specifically as follows: Judge the pixels of the picture. If the pixels of the picture are less than 2000×2000, directly perform a matting operation through the Visual Intelligence Open Platform to obtain the required main body picture of the commodity; if the pixels of the picture are greater than or equal to 2000×2000, use Imgproc.resize in OpenCV with the Imgproc.INTER_LANCZ0S4 algorithm to scale the picture proportionally to obtain a scaled picture; perform a matting operation on the scaled picture through the Visual Intelligence Open Platform to obtain a first mask picture; use Imgproc.resize in OpenCV with the Imgproc.INTER_LANCZ0S4 algorithm to scale the first mask picture proportionally back to the original size to obtain a second mask picture of the original size; read the alpha channel of each pixel point in the picture to obtain a first matrix, read the alpha channel of each pixel point in the second mask picture to obtain a second matrix, and call Core.min to merge the first matrix and the second matrix to obtain a third matrix; replace the alpha channel in the picture with the third matrix to obtain the required main body picture of the commodity; then convert the main body picture of the commodity into a main body mask picture of the commodity;

[0043] Due to the matting size limit of the Alibaba Cloud Visual Intelligence Open Platform, when the longest side exceeds 2000 pixels, proportional scaling is required. Call Imgproc.resize with the Imgproc.INTER_LANCZOS4 algorithm to perform proportional scaling of the picture;

[0044] Call Imgproc.resize with the Imgproc.INTER_LANCZOS4 algorithm to perform proportional scaling of the picture. This algorithm can reduce the influence of artifacts while maintaining the edge sharpness.

[0045] Obtain the matting result according to the black and white picture + the original picture:

[0046] a. Take out the alpha channel of the original picture. Call Core.min, which will take the minimum value of the transparency channels of the mask and the original picture, ensuring that the original information of the transparency channel is retained. Through this step of processing, the lines of the obtained matting can be made more perfect;

[0047] b. Remove the alpha channel of the original picture and add the black and white picture as the new alpha channel for layer merging;

[0048] c. Set the color value of the transparent area to black. This step is to reduce the size of the cutout result image and save storage costs.

[0049] The core code is as follows:

[0050] / / Create a fully transparent matrix of the same size as the original image for comparison;

[0051] Mat compareAlpha=new Mat(outImg.size(),CvType.CV_8UC1,Scala r.all(0.0));

[0052] / / Used to save the comparison results;

[0053] Mat compareResult=newMat();

[0054] / / Compare the transparent channel and alpha of the original image, and the position where the value is 0 in the obtained mask is the transparent position in the original image;

[0055] Core.compare(outPlanes.get(3),compareAlpha,compareResult,Core.CM P_EQ);

[0056] / / Create a completely black matrix of the same size as the original image;

[0057] Matblack=newMat(outImg.size(),outImg.type(),Scalar.all(0));

[0058] / / Copy the black in black to the corresponding position of outImg according to mask;

[0059] Core.bitwise_and(black,outImg,outImg,compareResult);

[0060] Through the above method, you can directly use the ready-made cutout software to cut out the image, and ensure that the quality of the image will not be reduced.

[0061] In this embodiment, preferably, step 3 is specifically as follows: If it is a pure product picture, the image-to-image algorithm calls the canny model to extract the edges of the product main body in the product main body mask picture, and sets the weight of canny to 0.85; the image-to-image algorithm calls the ipadapter algorithm to obtain the element features in the pure product picture, and sets the weight of ipadapter to 0.7; the product main body mask picture and the pure product picture are input into the image-to-image algorithm for mask redrawing to obtain a generated picture; the redrawing amplitude of the mask redrawing is set to 0.75;

[0062] If it is a product picture with a person, the image-to-image algorithm calls the canny model to extract the lines of the product main body and the person in the product picture with a person, and sets the canny weight to 0.5; the image-to-image algorithm calls the ipadapter algorithm and sets its weight to 0.9; the image-to-image algorithm calls the tile algorithm and sets its weight to 0.6; the product main body mask picture and the pure product picture are input into the image-to-image algorithm for mask redrawing to obtain a generated picture; the redrawing amplitude of the mask redrawing is set to 0.6;

[0063] In the above way, it can be ensured that the pattern of the product main body part will not change.

[0064] Based on the same inventive concept, the present application also provides an apparatus corresponding to the method in Embodiment 1. For details, see Embodiment 2.

[0065] Embodiment 2

[0066] As Figure 2 shown, in this embodiment, an apparatus for product material fission of product images is provided, including:

[0067] A picture classification module that judges the acquired pictures and classifies them into pure product pictures or product pictures with a person;

[0068] A product main body mask module that obtains the corresponding product main body mask picture through the picture;

[0069] A fission generation module. If it is a pure product picture, the image-to-image algorithm calls the canny model to extract the edges of the product main body in the product main body mask picture; the image-to-image algorithm calls the ipadapter algorithm to obtain the element features in the pure product picture; the product main body mask picture and the pure product picture are input into the image-to-image algorithm for mask redrawing to obtain a generated picture;

[0070] If it is a product picture with a person, the image generation algorithm calls the canny model to extract the lines of the product main body and the person in the product picture with a person; the image generation algorithm calls the ipadapter algorithm; the image generation algorithm calls the tile algorithm; the product main body mask picture and the pure product picture are input into the image generation algorithm for mask redrawing to obtain a generated picture.

[0071] In this embodiment, preferably, the picture classification module is specifically: using the openpose model to judge the acquired picture. The openpose model performs joint point detection on the picture. If the picture contains a person or a person's limb, it is a product picture with a person; otherwise, it is a pure product picture.

[0072] In this embodiment, preferably, the product main body mask module is specifically: judging the pixels of the picture. If the pixels of the picture are less than 2000×2000, directly perform a matte extraction operation through the Visual Intelligence Open Platform to obtain the required product main body picture; if the pixels of the picture are greater than or equal to 2000×2000, use Imgproc.resize in OpenCV with the Imgproc.INTER_LANCZ0S4 algorithm to scale the picture proportionally to obtain a scaled picture; perform a matte extraction operation on the scaled picture through the Visual Intelligence Open Platform to obtain a first mask picture; use Imgproc.resize in OpenCV with the Imgproc.INTER_LANCZ0S4 algorithm to scale the first mask picture proportionally back to the original size to obtain a second mask picture; read the alpha channel of each pixel point in the picture to obtain a first matrix, read the alpha channel of each pixel point in the second mask picture to obtain a second matrix, and call Core.min to merge the first matrix and the second matrix to obtain a third matrix; replace the alpha channel in the picture with the third matrix to obtain the required product main body picture; then convert the product main body picture into a product main body mask picture.

[0073] In this embodiment, preferably, the fission generation module is specifically: if it is a pure product picture, the image generation algorithm calls the canny model to extract the edges of the product main body in the product main body mask picture, and set the weight of canny to 0.85; the image generation algorithm calls the ipadapter algorithm to obtain the element features in the pure product picture, and set the weight of ipadapter to 0.7; input the product main body mask picture and the pure product picture into the image generation algorithm for mask redrawing to obtain a generated picture; set the redrawing amplitude of the mask redrawing to 0.75;

[0074] If it is a product picture with a person, the image - to - image generation algorithm calls the canny model to extract the lines of the product main body and the person in the product picture with a person, and sets the canny weight to 0.5; the image - to - image generation algorithm calls the ipadapter algorithm and sets its weight to 0.9; the image - to - image generation algorithm calls the tile algorithm and sets its weight to 0.6; the product main body mask picture and the pure product picture are input into the image - to - image generation algorithm for mask redrawing to obtain a generated picture; the redrawing amplitude of the mask redrawing is set to 0.6.

[0075] Since the device introduced in the second embodiment of the present invention is the device adopted for implementing the method of the first embodiment of the present invention, based on the method introduced in the first embodiment of the present invention, those skilled in the art can understand the specific structure and variations of the device, so it will not be elaborated here. Any device adopted for the method of the first embodiment of the present invention belongs to the scope protected by the present invention.

[0076] Based on the same inventive concept, this application provides an electronic device embodiment corresponding to the first embodiment. For details, see the third embodiment.

[0077] Embodiment Three

[0078] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, any implementation manner in the first embodiment can be realized.

[0079] Since the electronic device introduced in this embodiment is the device adopted for implementing the method in the first embodiment of this application, based on the method introduced in the first embodiment of this application, those skilled in the art can understand the specific implementation manner and various variations of the electronic device in this embodiment. Therefore, how the electronic device realizes the method in the embodiments of this application will not be introduced in detail here. Any device adopted by those skilled in the art for implementing the method in the embodiments of this application belongs to the scope protected by this application.

[0080] Based on the same inventive concept, this application provides a storage medium corresponding to the first embodiment. For details, see the fourth embodiment.

[0081] Embodiment Four

[0082] This embodiment provides a computer - readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, any implementation manner in the first embodiment can be realized.

[0083] The technical solutions provided in the embodiments of this application at least have the following technical effects or advantages:

[0084] In this embodiment, by quickly generating product background images with diverse styles, the efficiency of the design work is significantly improved. It not only ensures the coordination and consistency among the images but also realizes the natural integration between the background and the product. The pictures generated by this technical solution bring rich design inspiration and material references to designers and also lay a solid foundation for building a high-quality product image Lora model training library.

[0085] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0086] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0087] These computer program instructions can also be stored in a computer-readable memory capable of guiding a computer or other programmable data processing devices to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means realizes the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0088] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, so that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable devices provide steps for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0089] Although the specific embodiments of the present invention have been described above, those skilled in the art should understand that the specific embodiments we described are illustrative only and not used to limit the scope of the present invention. Equivalent modifications and changes made by those skilled in the art in accordance with the spirit of the present invention should be covered by the scope protected by the claims of the present invention.

Claims

1. A method for product material fission of a product image, characterized by: The steps include: Step 1: The obtained pictures are judged and classified into pure product pictures or product pictures with people; Step 2: Obtain the corresponding product main body mask image through the image; Step 3: If it is a pure product image, the image generation algorithm calls the Canny model to extract the edge of the product body of the product body mask image; the image generation algorithm calls the iPadapter algorithm to obtain the element features in the pure product image; the product body mask image and the pure product image are passed into the image generation algorithm for mask redrawing to obtain a generated image; If it is a product image with people, the image generation algorithm calls the Canny model to extract lines from the product body and the people in the product image with people; the image generation algorithm calls the iPadapter algorithm; the image generation algorithm calls the Tile algorithm; the product body mask image and the pure product image are passed into the image generation algorithm for mask redrawing to obtain the generated image.

2. The method for performing commodity material fission on a commodity image according to claim 1, characterized in that: The step 1 is specifically as follows: the acquired image is judged by the openpose model, and the openpose model detects joint points of the image. If the image contains a person or a person's limbs, it is a product image with a person; otherwise, it is a pure product image.

3. The method for performing commodity material fission on a commodity image according to claim 1, characterized in that: The step 2 is specifically as follows: judging the pixels of the image, if the pixels of the image are less than 2000×2000, directly performing a cutout operation through the visual intelligence open platform to obtain the required main body image of the product; if the pixels of the image are greater than or equal to 2000×2000, using Imgproc.resize in OpenCV and using the Imgproc.INTER_LANCZ0S4 algorithm to scale the image in proportion to obtain a scaled image; performing a cutout operation on the scaled image through the visual intelligence open platform to obtain a first mask image; using the Imgproc.resize in OpenCV and using the Imgproc.INTER_LANCZ0S4 algorithm to proportionally restore the image to a second mask image of the original size; reading out the transparent channel of each pixel in the image to obtain a first matrix, reading out the transparent channel of each pixel in the second mask image to obtain a second matrix, calling Core.min to merge the first matrix with the second matrix to obtain a third matrix; The transparent channel in the image is replaced by the third matrix to obtain the desired product main body image; and then the product main body image is converted into a product main body mask image.

4. The method for performing commodity material fission on a commodity image according to claim 1, characterized in that: The step 3 is specifically as follows: if it is a pure product image, the image generation algorithm calls the canny model to extract the edge of the product body of the product body mask image, and the weight of canny is set to 0.85; the image generation algorithm calls the ipadapter algorithm to obtain the element features in the pure product image, and the weight of ipadapter is set to 0.7; the product body mask image and the pure product image are passed into the image generation algorithm for mask redrawing to obtain a generated image; the redrawing amplitude of the mask redrawing is set to 0.75; If it is a product image with people, the image generation algorithm calls the canny model to extract lines from the product body and the people in the product image with people, and sets the canny weight to 0.5; the image generation algorithm calls the ipadapter algorithm and sets its weight to 0.9; the image generation algorithm calls the tile algorithm and sets its weight to 0.6; the product body mask image and the pure product image are passed into the image generation algorithm for mask redrawing to obtain a generated image; the redrawing amplitude of the mask redrawing is set to 0.

6.

5. A device for performing product material fission on a product image, characterized in that: include: The image classification module classifies the acquired images into pure product images or product images with people. The product body mask module obtains the corresponding product body mask image through the image; In the fission generation module, if it is a pure product image, the image generation algorithm calls the canny model to extract the edge of the product body of the product body mask image; the image generation algorithm calls the ipadapter algorithm to obtain the element features in the pure product image; the product body mask image and the pure product image are passed into the image generation algorithm for mask redrawing to obtain the generated image; If it is a product image with people, the image generation algorithm calls the Canny model to extract lines from the product body and the people in the product image with people; the image generation algorithm calls the iPadapter algorithm; the image generation algorithm calls the Tile algorithm; the product body mask image and the pure product image are passed into the image generation algorithm for mask redrawing to obtain the generated image.

6. The device for performing commodity material fission on commodity images according to claim 5, characterized in that: The image classification module specifically comprises: judging the acquired image through the openpose model, and detecting the joint points of the image. If the image contains a person or a person's limbs, it is a product image with a person; otherwise, it is a pure product image.

7. The device for performing commodity material fission on commodity images according to claim 5, characterized in that: The product main body mask module is specifically as follows: the pixels of the picture are judged. If the pixels of the picture are less than 2000×2000, a cutout operation is directly performed through the visual intelligence open platform to obtain the required product main body picture; if the pixels of the picture are greater than or equal to 2000×2000, the picture is scaled in proportion using the Imgproc.resize in OpenCV and the Imgproc.INTER_LANCZ0S4 algorithm to obtain a scaled picture; the scaled picture is cutout through the visual intelligence open platform to obtain a first mask picture; the first mask picture is restored to a second mask picture of the original size through the Imgproc.resize in OpenCV and the Imgproc.INTER_LANCZ0S4 algorithm; the transparent channel of each pixel in the picture is read out to obtain a first matrix, the transparent channel of each pixel in the second mask picture is read out to obtain a second matrix, and Core.min is called to merge the first matrix with the second matrix to obtain a third matrix; The transparent channel in the image is replaced by the third matrix to obtain the desired product main body image; and then the product main body image is converted into a product main body mask image.

8. The device for performing commodity material fission on commodity images according to claim 5, characterized in that: The fission generation module is specifically as follows: if it is a pure product image, the image generation algorithm calls the canny model to extract the edge of the product body of the product body mask image, and the weight of canny is set to 0.85; the image generation algorithm calls the ipadapter algorithm to obtain the element features in the pure product image, and the weight of ipadapter is set to 0.7; the product body mask image and the pure product image are passed into the image generation algorithm for mask redrawing to obtain a generated image; the redrawing amplitude of the mask redrawing is set to 0.75; If it is a product image with people, the image generation algorithm calls the canny model to extract lines from the product body and the people in the product image with people, and sets the canny weight to 0.5; the image generation algorithm calls the ipadapter algorithm and sets its weight to 0.9; the image generation algorithm calls the tile algorithm and sets its weight to 0.6; the product body mask image and the pure product image are passed into the image generation algorithm for mask redrawing to obtain a generated image; the redrawing amplitude of the mask redrawing is set to 0.

6.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 4 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images