Railway vehicle appearance image generation method and device based on deep learning

Abstract

The invention provides a rail vehicle appearance image generation method and device based on deep learning, and the method comprises the steps: inputting a first text description into a deep learning model, and obtaining a first image outputted by the deep learning model; the first text description comprises description of the expected railway vehicle appearance image by the user; under the condition that it is determined that the first image does not conform to the user expectation, inputting the first image and the second text description into a deep learning model, and obtaining a second image output by the deep learning model; the second text description comprises description of key points that the user expects to optimize the first graph. According to the deep learning-based railway vehicle appearance image generation method and device provided by the invention, the artificial intelligence technology is utilized to assist the design of the railway vehicle appearance image, and the image design efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of rail transit design, and in particular to a method and device for generating rail vehicle appearance images based on deep learning. Background Art

[0002] With the rapid development of urban rail transit construction, the status of rail transit vehicles in the modern urban transportation system has become increasingly prominent. Against this backdrop, the demand for rail transit vehicle design is showing an explosive growth trend. On the one hand, the market has put forward higher requirements for design efficiency, and it is necessary to launch novel and practical design solutions in a shorter time to adapt to the rapidly changing market environment and fierce competition. On the other hand, innovation has become a key indicator for measuring the quality of rail transit vehicle design. At the same time, the pursuit of excellence in detail processing is also indispensable, from the smoothness of the vehicle's lines to the coordination of color matching, from the clarity and eye-catching of the logo to the texture of the interior materials, every detail is related to the success or failure of the overall design.

[0003] Traditional vehicle design methods have gradually revealed their limitations in the face of these new challenges. Traditional design often relies too much on cumbersome manual drawing and complex 3D modeling, which not only takes a lot of time and manpower, but is also extremely inconvenient when modifying and adjusting design plans. Designers often need to spend a lot of time repeatedly drawing and modifying drawings. Once there is a change in requirements, the entire design process may need to start all over again, and the design efficiency is low, which makes it difficult for traditional design methods to quickly respond to market changes and meet the diverse needs of customers. Summary of the invention

[0004] The present invention provides a method and device for generating a rail vehicle appearance image based on deep learning, so as to solve the defect of low efficiency in designing the rail vehicle appearance image in the prior art and realize efficient generation of the rail vehicle appearance image.

[0005] The present invention provides a method for generating a rail vehicle appearance image based on deep learning, comprising: Inputting a first text description into a deep learning model to obtain a first image output by the deep learning model; wherein the first text description includes a user's description of a desired rail vehicle appearance image; When it is determined that the first image does not meet the user's expectations, the first image and the second text description are input into a deep learning model to obtain a second image output by the deep learning model; the second text description includes a description of key points that the user expects to optimize the first graphic; The deep learning model is obtained by training using rail vehicle appearance image samples based on a general large model.

[0006] In some embodiments, while inputting the first text description into the deep learning model, and / or while inputting the first image and the second text description into the deep learning model, the method further includes: An original source file is input into the deep learning model, wherein the original source file contains basic image materials related to generating an appearance image of a rail vehicle.

[0007] In some embodiments, the user's description of the desired rail vehicle appearance image includes at least one of the following information: Graphic style; Image format; Image size; The angle of the rail vehicle in the image; Shape features of rail vehicles in the image.

[0008] In some embodiments, the method further comprises: When it is determined that the second image does not meet user expectations, the second image and the third text description are input into a deep learning model to obtain an optimized second image output by the deep learning model; the third text description includes a description of the key points that the user expects to optimize the second graphic.

[0009] In some embodiments, the deep learning model that outputs the first image and the deep learning model that outputs the second image are the same model.

[0010] In some embodiments, the deep learning model that outputs the first image and the deep learning model that outputs the second image are two different models.

[0011] The present invention also provides a rail vehicle appearance image generation device based on deep learning, characterized in that it includes: A first generating module is used to input a first text description into a deep learning model to obtain a first image output by the deep learning model; the first text description includes a user's description of a desired rail vehicle appearance image; A second generating module is used for inputting the first image and a second text description into a deep learning model to obtain a second image output by the deep learning model when it is determined that the first image does not meet the user's expectations; the second text description includes a description of key points that the user expects to optimize the first graphic; The deep learning model is obtained by training using rail vehicle appearance image samples based on a general large model.

[0012] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for generating an appearance image of a rail vehicle based on deep learning as described above is implemented.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for generating a rail vehicle appearance image based on deep learning.

[0014] The present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-described methods for generating a rail vehicle appearance image based on deep learning.

[0015] The method and device for generating a rail vehicle appearance image based on deep learning provided by the present invention utilize artificial intelligence technology to generate a design of a rail vehicle appearance image, thereby improving the efficiency of image design. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 It is a flowchart of the method for generating rail vehicle appearance images based on deep learning provided by the present invention.

[0018] Figure 2 is a schematic diagram of a rail vehicle appearance image generation system based on deep learning provided by the present invention.

[0019] Figure 3 It is a schematic diagram of image changes during the image generation process provided by the present invention.

[0020] Figure 4 It is a structural schematic diagram of a rail vehicle appearance image generation device based on deep learning provided by the present invention.

[0021] Figure 5 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] Figure 1 is a flow chart of a method for generating a rail vehicle appearance image based on deep learning provided by the present invention, such as Figure 1 As shown, the method includes the following: Step 101: input a first text description into a deep learning model to obtain a first image output by the deep learning model; the first text description includes a user's description of a desired rail vehicle appearance image.

[0024] In some embodiments, the text description can be directly input into the deep learning model to obtain a first image output by the deep learning model.

[0025] The first text description includes a user's description of the desired rail vehicle appearance image.

[0026] In some embodiments, the user's description of the desired rail vehicle appearance image includes at least one of the following information: Graphic style; Image format; Image size; The angle of the rail vehicle in the image; Shape features of rail vehicles in the image.

[0027] In some embodiments, while inputting the first text description into the deep learning model, and / or while inputting the first image and the second text description into the deep learning model, the method further includes: An original source file is input into the deep learning model, wherein the original source file contains basic image materials related to generating an appearance image of a rail vehicle.

[0028] These basic image materials include sketches, component drawings, etc.

[0029] The deep learning model is trained based on the general large model using rail vehicle appearance image samples.

[0030] The number of first images output by the deep learning model may be one or more.

[0031] Step 102: if it is determined that the first image does not meet the user's expectations, input the first image and the second text description into a deep learning model to obtain a second image output by the deep learning model; the second text description includes a description of the key points that the user expects to optimize the first graphic; The deep learning model is obtained by training using rail vehicle appearance image samples based on a general large model.

[0032] Specifically, the first image of the deep learning model may or may not meet the user's expectations.

[0033] When it is determined that the first image meets the user's expectations, the first image and the second text description are no longer input into the deep learning model, and the first image is directly output.

[0034] When it is determined that the first image does not meet the user's expectations, the first image and the second text description are input into the deep learning model to obtain the second image output by the deep learning model; the second text description includes a description of the key points that the user expects to optimize the first graphic.

[0035] The key points of optimization include one or more of image coloring, image refinement, inspiration fusion, and local redrawing.

[0036] In some embodiments, when it is determined that the second image does not meet user expectations, the second image and the third text description are input into a deep learning model to obtain an optimized second image output by the deep learning model; the third text description includes a description of the key points that the user expects to optimize the second graphic.

[0037] Furthermore, the above steps may be repeated until the optimization of the second image meets the user's expectations.

[0038] In some embodiments, the deep learning model that outputs the first image and the deep learning model that outputs the second image are the same model.

[0039] In some embodiments, the deep learning model that outputs the first image and the deep learning model that outputs the second image are two different models.

[0040] The rail vehicle appearance image generation method based on deep learning provided by the present invention utilizes artificial intelligence technology to generate the design of the rail vehicle appearance image, thereby improving the efficiency of image design.

[0041] Figure 2 is a schematic diagram of a rail vehicle appearance image generation system based on deep learning provided by the present invention, such as Figure 2As shown, the rail vehicle appearance image generation system based on deep learning includes an input module, a model training module, an image editing and optimization module, and an output module.

[0042] The input module is used to obtain the original file source for generating the user's target image. The original file source is the image material basis for image generation. The input module includes a text unit and a picture unit. The text unit is used to input the first text description, and generate an image file (first image) matching the text content for the first text description through a preset large model (a trained deep learning model), and save the generated image file in a local file path or a network path. The picture unit is used to obtain an image file from a local file path or a network path, and read the image file. It is displayed through the image display component of the image editing and optimization module.

[0043] The model training module is used to train the deep learning model. For example, the deep learning model parameters are trained by adjusting the parameters and the deep learning model parameters are saved.

[0044] In some embodiments, when the user inputs the first text description through the text unit, the system processes it according to the settings of the text map module. The forward / reverse description word settings are retained, and the Chinese-English translator can be involved through secondary development to ensure the accuracy of the description.

[0045] Figure 3 is a schematic diagram of image changes during the image generation process provided by the present invention, such as Figure 3 As shown, the original file source input to the deep learning model contains a sketch of a rail vehicle (such as Figure 3 In (a) of the figure, and also input a first text description, for example, the first text description is "high-speed EMU appearance, front 45-degree viewing angle, white body, blue ribbon, simple style", the system will recognize and process these descriptive words and output multiple generated images (such as Figure 3 (b) in the figure.

[0046] In this embodiment, during model training, the deep learning model is fine-tuned and trained according to the settings of the model training module. When adjusting parameter training, reference is made to the relevant parameter setting method for model training, such as learning rate, number of iterations, etc., and the trained parameters are saved.

[0047] The model training module selection is set as two selection boxes, one for selecting the deep learning model imported into the software, and the other for setting the weight value parameter (0-100%). For example, if a specific rail vehicle detail enhancement Lora model is selected, the appropriate weight can be set to adjust its influence on the generated image.

[0048] If for Figure 3 If the images generated by (b) meet the user's expectations, they are directly output (such as Figure 3 (c) in the figure.

[0049] If none of the generated images meet the user's expectations, then images with optimization potential can be selected (such as Figure 3 (d) in the figure is further optimized.

[0050] The image editing and optimization module is used to edit and further optimize the generated first image. The user issues editing and optimization instructions to the image through the image editing control and the image optimization control, such as Figure 3 In (d), mark the selected images (such as Figure 3 The yellow area in (d) is selected, and the marked area is optimized; the image editing control includes a size adjustment control and an image mode selection control. The size adjustment control is used to adjust the length and width of the image file by dragging the control; the image mode selection control is used to adjust the style mode of the current image file, and display the modified preview image through the image editing and optimization module; the image optimization control includes an image coloring unit, an image refinement unit, an inspiration fusion unit, and a local redrawing unit. The image optimization control is used to achieve different optimization effects through different units.

[0051] The image editing and optimization module uses graphic parameterization and numerical parameterization to adjust the parameters of the image editing control and the image optimization control. Graphic parameterization is used to adjust the parameters of the image editing control and the image optimization control through preset visual graphic parameter adjustment pages. Each visual graphic parameter adjustment page corresponds to a set of preset parameters. The preset parameters and image files are read through the deep learning model, and the image is generated and adjusted according to the preset parameters; numerical parameterization is used for users to self-adjust the image generation parameters. For each parameter, the user's self-set value is input to generate a parameter configuration group. After the deep learning model adjusts the image through array parameterized parameters, the parameters of the parameter configuration group can be saved to form preset parameters for use on the graphic parameter adjustment page; the image mode selection control includes a graphic parameterization option for the image generation style. The image mode selection control presets preset parameter configuration groups for different style adjustments. After the deep learning model receives the parameters of the parameter configuration group, it sets the style conversion from the current style to the parameter configuration group.

[0052] The image coloring unit of the image optimization control is used to color the line drawing mode under the image mode selection control. The user can automatically complete the coloring work through text description or by clicking on the image of the line drawing mode; the image refinement unit is used for the user to refine the parameters of the image file. The user selects the refined area on the image file and enters the text description of the refined adjustment. The deep learning model generates the image on the selected area according to the text description; the inspiration fusion unit is used to fuse the image material and text description provided by the user to the image file through the deep learning model, and generate a style image similar to the image material and text description; the local redrawing unit is used to regenerate the image of the local area of ​​the image selected by the user, and the generation parameters are determined by the user entering the text description or selecting the preset redrawing parameters.

[0053] In this embodiment, when adjusting the style mode of the current image file through the image model selection control, the modes of the image coloring and image refinement modules in the image generation module are set. For example, in the image coloring module, modes such as line draft and white model are set, and different parameter configuration groups are preset for each mode, and adjusted through graphic parameterization and numerical parameterization.

[0054] After the deep learning model edits and further optimizes the image, one or more images (the second image, such as Figure 3 (e) in the figure.

[0055] If for Figure 3 If the images generated by (e) meet the user's expectations, they are directly output (such as Figure 3 (f) in the figure.

[0056] If the generated images do not meet the user's expectations, the above optimization steps can be repeated until they meet the user's expectations.

[0057] The present invention provides a method that can automatically generate rail vehicle appearance images that meet the requirements based on user input parameters or keywords, and can support users to perform subsequent editing and optimization, thereby greatly improving design efficiency and creativity. By flexibly configuring the deep learning model generated by the image, the efficiency of image generation and the convenience of model replacement are improved.

[0058] The rail vehicle appearance image generation device based on deep learning provided by the present invention is described below. The rail vehicle appearance image generation device based on deep learning described below and the rail vehicle appearance image generation method based on deep learning described above can refer to each other.

[0059] Figure 4 is a schematic diagram of the structure of a rail vehicle appearance image generation device based on deep learning provided by the present invention, such as Figure 4As shown, the present invention provides a rail vehicle appearance image generation device based on deep learning, comprising: The first generating module 401 is used to input the first text description into the deep learning model to obtain the first image output by the deep learning model; the first text description includes the user's description of the desired rail vehicle appearance image; The second generation module 402 is used for inputting the first image and the second text description into the deep learning model to obtain the second image output by the deep learning model when it is determined that the first image does not meet the user's expectations; the second text description includes a description of the key points that the user expects to optimize the first graphic; The deep learning model is obtained by training using rail vehicle appearance image samples based on a general large model.

[0060] Specifically, the rail vehicle appearance image generation method based on deep learning provided in the embodiment of the present application can refer to the above-mentioned rail vehicle appearance image generation method embodiment based on deep learning, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the above-mentioned corresponding method embodiments will not be described in detail here.

[0061] Figure 5 is a schematic diagram of the structure of the electronic device provided by the present invention, such as Figure 5 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the rail vehicle appearance image generation method based on deep learning, and the method includes: Inputting a first text description into a deep learning model to obtain a first image output by the deep learning model; wherein the first text description includes a user's description of a desired rail vehicle appearance image; When it is determined that the first image does not meet the user's expectations, the first image and the second text description are input into a deep learning model to obtain a second image output by the deep learning model; the second text description includes a description of key points that the user expects to optimize the first graphic; The deep learning model is obtained by training using rail vehicle appearance image samples based on a general large model.

[0062] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0063] On the other hand, the present invention further provides a computer program product, the computer program product comprising a computer program, the computer program can be stored on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, the computer can execute the rail vehicle appearance image generation method based on deep learning provided by the above methods, the method comprising: Inputting a first text description into a deep learning model to obtain a first image output by the deep learning model; wherein the first text description includes a user's description of a desired rail vehicle appearance image; When it is determined that the first image does not meet the user's expectations, the first image and the second text description are input into a deep learning model to obtain a second image output by the deep learning model; the second text description includes a description of key points that the user expects to optimize the first graphic; The deep learning model is obtained by training using rail vehicle appearance image samples based on a general large model.

[0064] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to execute the rail vehicle appearance image generation method based on deep learning provided by the above methods, the method comprising: Inputting a first text description into a deep learning model to obtain a first image output by the deep learning model; wherein the first text description includes a user's description of a desired rail vehicle appearance image; When it is determined that the first image does not meet the user's expectations, the first image and the second text description are input into a deep learning model to obtain a second image output by the deep learning model; the second text description includes a description of key points that the user expects to optimize the first graphic; The deep learning model is obtained by training using rail vehicle appearance image samples based on a general large model.

[0065] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0066] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0067] It should also be noted that the terms "target", "first", "second", etc. in the embodiments of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first" and "second" are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more.

[0068] In the embodiments of the present application, the term "and / or" describes the association relationship of the associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0069] "Determine B based on A" in the embodiments of the present application means that the factor A should be considered when determining B. It is not limited to "B can be determined based on A alone", but should also include: "Determine B based on A and C", "Determine B based on A, C and E", "Determine C based on A, and further determine B based on C", etc. In addition, it can also include taking A as a condition for determining B, for example, "When A meets the first condition, use the first method to determine B"; for another example, "When A meets the second condition, determine B", etc.; for another example, "When A meets the third condition, determine B based on the first parameter", etc. Of course, it can also be a condition that takes A as a factor for determining B, for example, "When A meets the first condition, use the first method to determine C, and further determine B based on C", etc.

[0070] In the embodiments of the present application, the term "plurality" refers to two or more than two, and other quantifiers are similar.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating rail vehicle appearance images based on deep learning, characterized in that: include: Inputting a first text description into a deep learning model, and obtaining a first image output by the deep learning model; The first text description includes a user's description of the desired rail vehicle appearance image; When it is determined that the first image does not meet the user's expectations, the first image and the second text description are input into a deep learning model to obtain a second image output by the deep learning model; the second text description includes a description of key points that the user expects to optimize the first graphic; The deep learning model is obtained by training using rail vehicle appearance image samples based on a general large model.

2. The method for generating rail vehicle appearance images based on deep learning according to claim 1, characterized in that: While inputting the first text description into the deep learning model, and / or while inputting the first image and the second text description into the deep learning model, the method further includes: An original source file is input into the deep learning model, wherein the original source file contains basic image materials related to generating an appearance image of a rail vehicle.

3. The method for generating rail vehicle appearance images based on deep learning according to claim 1, characterized in that: The user's description of the desired rail vehicle appearance image includes at least one of the following information: Graphic style; Image format; Image size; The angle of the rail vehicle in the image; Shape features of rail vehicles in the image.

4. The method for generating rail vehicle appearance images based on deep learning according to claim 1, characterized in that: The method further comprises: When it is determined that the second image does not meet user expectations, the second image and the third text description are input into a deep learning model to obtain an optimized second image output by the deep learning model; the third text description includes a description of the key points that the user expects to optimize the second graphic.

5. The method for generating rail vehicle appearance images based on deep learning according to claim 1, characterized in that: The deep learning model that outputs the first image and the deep learning model that outputs the second image are the same model.

6. The method for generating rail vehicle appearance images based on deep learning according to claim 1, characterized in that: The deep learning model that outputs the first image and the deep learning model that outputs the second image are two different models.

7. A rail vehicle appearance image generation device based on deep learning, characterized in that: include: A first generating module is used to input a first text description into a deep learning model to obtain a first image output by the deep learning model; the first text description includes a user's description of a desired rail vehicle appearance image; A second generating module is used for inputting the first image and a second text description into a deep learning model to obtain a second image output by the deep learning model when it is determined that the first image does not meet the user's expectations; the second text description includes a description of key points that the user expects to optimize the first graphic; The deep learning model is obtained by training using rail vehicle appearance image samples based on a general large model.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for generating a rail vehicle appearance image based on deep learning as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for generating a rail vehicle appearance image based on deep learning as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for generating a rail vehicle appearance image based on deep learning as described in any one of claims 1 to 6 is implemented.

Images