Artificial intelligence-based blank house decoration effect picture generation method and related device

Abstract

The invention is suitable for the technical field of internet decoration, provides a blank house decoration effect generation method based on artificial intelligence and a related device, and realizes that a corresponding target decoration effect picture is quickly obtained according to a target blank house picture by utilizing an artificial intelligence technology. The method mainly comprises the following steps: acquiring a target blank house picture; performing deep processing on the target workblank room picture by using a first preset artificial intelligence model to obtain a target workblank room depth evaluation map; segmenting the target workblank house picture by using a second preset artificial intelligence model to obtain a target workblank house segmentation map; inputting the target workblank room depth evaluation map and the target workblank room segmentation map into a third preset artificial intelligence model, and outputting a picture constraint of the target workblank room picture; and generating a target decoration effect picture according to the target blank house picture, the picture constraint and an input preset creation prompt word.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of Internet decoration, and particularly relates to a raw house decoration effect generation method based on artificial intelligence and a related device. BACKGROUND

[0002] A raw house, also known as a primary decoration house, is usually a house with only basic treatment and no surface treatment. In the current real estate market, raw houses still account for an important proportion of house transactions. For a homeowner who has purchased a raw house, before choosing to decorate the raw house, the homeowner often hopes that a decoration design company can provide a decoration effect drawing of different decoration styles for the house type of the raw house, so that the homeowner can see the decoration effect drawing of the house type of the raw house purchased by the homeowner before decoration, which is beneficial for the homeowner to make a choice on different decoration styles and decoration details of the house type of the raw house purchased by the homeowner.

[0003] However, the current generation of decoration effect drawings of raw houses by decoration design companies needs to arrange engineers to go to the houses to obtain house data, use professional software to design models, and finally render the designed models to obtain decoration effect drawings. This process is time-consuming and not friendly to the homeowner experience. SUMMARY

[0004] The purpose of the present application is to provide a raw house decoration effect generation method based on artificial intelligence and a related device, which realizes the use of artificial intelligence technology to quickly obtain a corresponding target decoration effect drawing according to a target raw house picture.

[0005] In a first aspect, the present application provides a raw house decoration effect generation method based on artificial intelligence, comprising:

[0006] obtaining a target raw house picture;

[0007] performing depth processing on the target raw house picture by using a first preset artificial intelligence model to obtain a target raw house depth evaluation drawing, the first preset artificial intelligence model being a trained model capable of identifying the depth of each region of a building raw picture;

[0008] performing segmentation on the target raw house picture by using a second preset artificial intelligence model to obtain a target raw house segmentation drawing, the second preset artificial intelligence model being a trained model capable of identifying the type of each building structure of a building raw picture;

[0009] inputting the target raw house depth evaluation drawing and the target raw house segmentation drawing into a third preset artificial intelligence model to output a picture constraint of the target raw house picture, the third preset artificial intelligence model being a trained model capable of generating a picture constraint corresponding to a raw house picture according to the raw house depth evaluation drawing and the raw house segmentation drawing corresponding to the raw house picture.

[0010] generating a target decoration effect picture according to the target raw house picture, the picture constraint information, and an input preset creation prompt word.

[0011] Optionally, generating a target decoration effect picture according to the target raw house picture, the picture constraint information, and an input preset creation prompt word comprises:

[0012] inputting the target raw house picture, the picture constraint information, and the preset creation prompt word into a fourth preset artificial intelligence model to obtain a target decoration effect picture corresponding to the target raw house picture, the fourth preset artificial intelligence model being a trained model capable of generating a decoration effect picture corresponding to a raw house picture according to the raw house picture, picture constraint information corresponding to the raw house picture, and an input creation prompt word.

[0013] Optionally, the first preset artificial intelligence model is a depth_midas depth evaluation model.

[0014] Optionally, the second preset artificial intelligence model is a seg_ofade20k segmentation model.

[0015] Optionally, the third preset artificial intelligence model is a controlnet model.

[0016] Optionally, the fourth preset artificial intelligence model is a stablediffusion model.

[0017] In a second aspect, the present application provides a raw house decoration effect generation device based on artificial intelligence, comprising:

[0018] an acquisition unit configured to acquire a target raw house picture;

[0019] a processing unit configured to perform depth processing on the target raw house picture by using a first preset artificial intelligence model to obtain a target raw house depth evaluation picture, the first preset artificial intelligence model being a trained model capable of identifying the depth of each region of a building raw picture.

[0020] a segmentation unit configured to perform segmentation on the target raw house picture by using a second preset artificial intelligence model to obtain a target raw house segmentation picture, the second preset artificial intelligence model being a trained model capable of identifying the type of each building structure of a building raw picture.

[0021] A constraint unit is configured to input the target rough house depth evaluation map and the target rough house segmentation map into a third preset artificial intelligence model, and output picture constraints of the target rough house picture, wherein the third preset artificial intelligence model is a trained model capable of generating picture constraints corresponding to a rough house picture according to a rough house depth evaluation map and a rough house segmentation map corresponding to the rough house picture.

[0022] A generation unit is configured to generate a target decoration effect picture according to the target rough house picture, the picture constraints, and a preset creative prompt word input.

[0023] Optionally, when the generation unit generates the target decoration effect picture according to the target rough house picture, the picture constraint information, and the preset creative prompt word input, the generation unit is specifically configured to:

[0024] input the target rough house picture, the picture constraint information, and the preset creative prompt word into a fourth preset artificial intelligence model to obtain a target decoration effect picture corresponding to the target rough house picture, wherein the fourth preset artificial intelligence model is a trained model capable of generating a decoration effect picture corresponding to a rough house picture according to the rough house picture, picture constraint information corresponding to the rough house picture, and a creative prompt word input.

[0025] Optionally, the first preset artificial intelligence model is a depth_midas depth evaluation model.

[0026] Optionally, the second preset artificial intelligence model is a seg_ofade20k segmentation model.

[0027] Optionally, the third preset artificial intelligence model is a controlnet model.

[0028] Optionally, the fourth preset artificial intelligence model is a stablediffusion model.

[0029] In a third aspect, a computer device is provided, comprising:

[0030] a processor, a memory, a bus, an input / output interface, and a network interface;

[0031] the processor is connected to the memory, the input / output interface, and the network interface through the bus;

[0032] the memory stores a program;

[0033] when the processor executes the program stored in the memory, the method for generating a decoration effect picture of a rough house based on artificial intelligence in any one of the first aspect is implemented.

[0034] In a fourth aspect, the present application provides a computer storage medium, wherein the computer storage medium stores instructions, and the instructions cause a computer to execute the method for generating a decoration effect of a rough house based on artificial intelligence in any one of the first aspect.

[0035] In a fifth aspect, the present application provides a computer program product, wherein the computer program product causes a computer to execute the method for generating a decoration effect of a rough house based on artificial intelligence in any one of the first aspect when the computer program product is executed on the computer.

[0036] The above technical solutions can be seen that the embodiments of the present application have the following advantages:

[0037] The method for generating a decoration effect of a rough house based on artificial intelligence in the embodiment can obtain a target rough house picture, and then utilize a first preset artificial intelligence model to perform deep processing on the target rough house picture to obtain a target rough house depth evaluation picture, wherein the first preset artificial intelligence model is a trained model capable of identifying depths of each region of a building rough picture; and can utilize a second preset artificial intelligence model to segment the target rough house picture to obtain a target rough house segmentation picture, wherein the second preset artificial intelligence model is a trained model capable of identifying each building structure type of a building rough picture; then input the target rough house depth evaluation picture and the target rough house segmentation picture into a third preset artificial intelligence model to output a picture constraint of the target rough house picture, wherein the third preset artificial intelligence model is a trained model capable of generating a picture constraint corresponding to the rough house picture according to a rough house depth evaluation picture and a rough house segmentation picture corresponding to the rough house picture; and finally generate a target decoration effect picture according to the target rough house picture, the picture constraint, and an input preset creation prompt word, so as to realize the use of artificial intelligence technology to quickly obtain a corresponding target decoration effect picture according to a target rough house picture. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 An embodiment flowchart of the method for generating a decoration effect of a rough house based on artificial intelligence of the present application is shown in the figure;

[0039] Figure 2 Another embodiment flowchart of the method for generating a decoration effect of a rough house based on artificial intelligence of the present application is shown in the figure;

[0040] Figure 3 An embodiment structure diagram of the device for generating a decoration effect of a rough house based on artificial intelligence of the present application is shown in the figure;

[0041] Figure 4 An embodiment structure diagram of the computer device of the present application is shown in the figure;

[0042] Figure 5A schematic diagram of a target rough house image for applying an embodiment of the method for generating rough house decoration effects based on artificial intelligence of the present application;

[0043] Figure 6 for Figure 5 A schematic diagram of an embodiment of the effect of obtaining a depth assessment map of a target rough housing after deep processing by a first preset artificial intelligence model;

[0044] Figure 7 for Figure 5 A schematic diagram of an embodiment of the effect of obtaining a target rough house segmentation map after segmentation by a second preset artificial intelligence model;

[0045] Figure 8 This is a schematic diagram of an interface for generating a target decoration rendering based on a target rough house picture, picture constraints, and input preset creation prompts;

[0046] Figure 9 for Figure 8 A schematic diagram of an embodiment of a target decoration effect diagram generated corresponding to the embodiment;

[0047] Figure 10 This is a schematic diagram of an embodiment of drawing an image output by a line segment detection model of a ControlNet model according to line segments in this application;

[0048] Figure 11 The exhibition will Figure 6 The target rough housing depth assessment diagram shown is a schematic diagram of an embodiment of loading the ControlNet model as a reference diagram;

[0049] Figure 12 The exhibition will Figure 7 The target rough housing segmentation map shown is a schematic diagram of an embodiment of loading the target rough housing segmentation map into the ControlNet model as a reference map. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of this application more clearly understood, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0051] In the field of building decoration, generating a decoration effect drawing for a raw house for the reference of the owner has always been a prerequisite for the cooperation between the decoration design company and the owner. This is because the raw house is usually only a building framework without surface treatment, which cannot intuitively let the owner feel the space and style positioning after decoration, which brings many challenges to the owner's decoration decision. And through professional design tools to generate decoration effect drawing, it can convert abstract design concepts into visual images, providing important reference for the owner. At the same time, under the condition of cement wall and exposed pipeline of raw house, it is difficult for the owner to imagine the final decoration effect. If only relying on the verbal description or plan drawing of the engineer, it is easy for the owner to have understanding deviation. The decoration effect drawing in the prior art can intuitively display the space layout, color matching and furniture configuration through three-dimensional modeling, material rendering and light simulation, help the owner quickly understand the intention of the engineer, avoid repeated modification due to cognitive difference, and significantly improve the communication efficiency.

[0052] Although it is realized that the decoration effect drawing is an important communication link for the cooperation between the decoration design company and the owner, at present, the decoration design company needs to arrange engineers to measure the house to obtain house data, then use professional software to model and design, and finally render the model designed by modeling to obtain the decoration effect drawing. This process is time-consuming, and usually the decoration design company needs to design a variety of style effect drawings in advance for display, and it is possible that part or most of the decoration effect drawings adapted to the owner's raw house data modeling, design and rendering do not meet the owner's individual requirements and expectations, resulting in the decoration design company needs to pay a large amount of manpower and time cost in advance, and the experience of the owner is not friendly.

[0053] In view of this, this embodiment provides a rough house decoration effect generation solution based on artificial intelligence, which runs in a rough house decoration effect generation system based on artificial intelligence. The rough house decoration effect generation system based on artificial intelligence includes: a central processing unit module, a memory module, an artificial intelligence chip module, a communication module, a power supply module, etc., and is connected to a printed circuit board (PCB). The above modules are connected to a computer board (PCB) to obtain a so-called artificial intelligence-based rough house decoration effect generation system; wherein the central processing unit is mainly used to comprehensively dispatch the various component modules connected to it to implement the artificial intelligence-based rough house decoration effect generation method mentioned in this embodiment; the power module is mainly used to provide a suitable operating voltage to the central processing unit module, the memory module, the artificial intelligence chip module, the communication module, etc. The power module can be powered by a built-in battery or a mains power supply mode, and the power source of the power module is not limited here; the memory module mainly stores the program for implementing the relevant steps of this embodiment, and the memory also stores various preset trained artificial intelligence models, etc.; the artificial intelligence chip module is used to quickly load and run the artificial intelligence model stored in the memory module, and process the target rough house image and other images to obtain the rough house depth assessment map, rough house segmentation map, image constraints, decoration effect map, etc., which are stored in the memory module; the communication module is mainly used to communicate with external devices (mobile phones, tablets, computers and other electronic devices) (for example, obtaining the target rough house image, outputting the rough house depth assessment map, rough house segmentation map, image constraints, decoration effect map, etc.).

[0054] Based on the above understanding, please refer to 1. An embodiment of the method for generating the decoration effect of a rough house based on artificial intelligence in this application includes:

[0055] 101. Get the target rough house picture.

[0056] It is understandable that the target rough house picture obtained in this embodiment can be taken and uploaded by the owner himself, or by the staff of the decoration design company, for example Figure 5 As shown, the threshold for use is lowered. Anyone who registers and logs in to the artificial intelligence-based rough house decoration effect generation system of this embodiment to become a user can upload the target rough house picture according to the operating instructions; preferably, in order to make the target decoration effect picture generated by this embodiment more in line with the public's aesthetic taste and reflect the details of a functional area of ​​the rough house as much as possible, requirements are put forward for the shooting angle, clarity, picture format, resolution, etc. of the target rough house picture obtained in this step. The shooting requirements of the target rough house picture can be set according to actual needs and are not further limited here.

[0057] 102, using the first preset artificial intelligence model to perform deep processing on the target raw house picture to obtain a target raw house depth evaluation map.

[0058] The target raw house picture obtained in step 101 is input into the first preset artificial intelligence model to perform deep processing on the target raw house picture to obtain a target raw house depth evaluation map. For example, the target raw house picture in step 101 is input into the first preset artificial intelligence model to obtain a target raw house depth evaluation map as shown in FIG. 4. Figure 5 The target raw house picture is processed by the first preset artificial intelligence model to obtain a target raw house depth evaluation map, as shown in FIG. 4. Figure 6 The so-called depth in this embodiment refers to image depth, that is, the number of bits used to store each pixel; image depth refers to the number of bits required to store the gray scale or color of the pixel depth. Assuming that the pixel depth of the image is 16 bits, but the number of bits used to represent the gray scale or color of the image is only 15 bits, then the image depth of the image is 15. Image depth determines the number of possible colors or the number of possible gray levels for each pixel of the image. For example, a color image is represented by R, G, and B components, each component uses 8 bits, and the pixel depth is 24 bits.

[0059] The first preset artificial intelligence model in this embodiment is a trained model capable of identifying the depth of each region of the building raw map. The training of the first preset artificial intelligence model can collect a large number of raw house pictures as training samples, and manually label and classify each pixel corresponding to each raw house picture to train the artificial intelligence model to obtain a trained model capable of identifying the depth of each region of the building raw map. It can be understood that the first preset artificial intelligence model in this embodiment includes a control_v11f1p_sd15_depth[cfd03158] model based on controlnet, a depth_midas depth evaluation model, etc.

[0060] 103, using the second preset artificial intelligence model to segment the target raw house picture to obtain a target raw house segmentation map.

[0061] The target raw house picture obtained in step 101 is input into the second preset artificial intelligence model to perform segmentation on the target raw house picture to obtain a target raw house segmentation map. For example, the target raw house picture in step 101 is input into the second preset artificial intelligence model to obtain a target raw house segmentation map as shown in FIG. 5. Figure 5 The target raw house picture is processed by the second preset artificial intelligence model to obtain a target raw house segmentation map, as shown in FIG. 5. Figure 7The picture segmentation of the raw house in this embodiment refers to identifying and segmenting the regions of the same type (such as walls, windows, floors, ceilings, etc.) in the target raw house picture to clearly define the boundaries of these regions. The second preset artificial intelligence model in this embodiment is a trained model capable of identifying various building structure types of the building raw picture. The training of the above-mentioned second preset artificial intelligence model can collect a large number of raw house pictures as training samples, and combine manual labeling and classification of the functional regions corresponding to each raw house picture to train the artificial intelligence model, so as to obtain a trained model capable of segmenting the target raw house picture. It can be understood that the second preset artificial intelligence model in this embodiment includes a control_v11p_sd15_seg[e1f51eb9] model based on controlnet, a seg_ofade20k segmentation model, etc.

[0062] 104. Input the target raw house depth evaluation picture and the target raw house segmentation picture into the third preset artificial intelligence model, and output the picture constraint of the target raw house picture.

[0063] Input the target raw house depth evaluation picture obtained in step 102 and the target raw house segmentation picture obtained in step 103 into the third preset artificial intelligence model, and output the picture constraint of the target raw house picture. For example, input the target raw house depth evaluation picture and the target raw house segmentation picture into the third preset artificial intelligence model, and output the picture constraint of the target raw house picture as shown in FIG. 10. Figure 6 、 Figure 7 Input the third preset artificial intelligence model, and output the picture constraint of Figure 5 . The picture constraint in this embodiment includes but is not limited to edge detection, depth, semantic segmentation, etc., wherein the edge detection is used to control the generation of the image and is suitable for scenes requiring accurate edges; the depth information is used to control the generation of the image and is suitable for scenes requiring three-dimensional perception; the semantic segmentation is used to control the generation of the image and is suitable for image generation scenes requiring detailed region division. It can be understood that the third preset artificial intelligence model in this embodiment is a stable diffusion model.

[0064] 105. Generate a target decoration effect picture according to the target raw house picture, the picture constraint and the preset creative prompt word input.

[0065] Further, this step can generate a target decoration effect picture according to the target raw house picture obtained in step 101, the picture constraint obtained in step 104 and the preset creative prompt word input by the user directly or selected by the user; for example, the preset creative prompt word is that the space type is selected as a living room and the style type is selected as a light luxury style, etc., as shown in FIG. 11; and finally generate a target decoration effect picture as shown in the right side of FIG. 12, i.e. Figure 8 . Figure 8 Figure 9 .

[0066] ​It should be noted that the usage scenario of the method for generating the decoration effect of a rough house based on artificial intelligence in this embodiment is usually online (such as APP, Internet website, WeChat applet, etc.), and the owner or the engineer who comes to communicate with the owner can be directly contacted and go directly to the decoration site; in addition, in the preliminary communication of the plan with the owner, the owner in most cases only needs to see the decoration effect picture of the rough house from a specific angle, and does not need a panoramic view. This embodiment directly generates the target decoration effect picture based on the target rough house picture, which just meets the needs of this scenario. After the owner browses many types of target decoration effect pictures on site, he can currently choose the target decoration effect picture that he is most satisfied with, and then obtain the house data based on this, use professional software for modeling and design, and finally render the model designed to obtain the decoration effect picture, which is more efficient and more owner-friendly.

[0067] Please refer to 2, an embodiment of the method for generating the decoration effect of a rough house based on artificial intelligence in this application, including:

[0068] 201. Obtain the target rough house picture.

[0069] The execution of this step is the same as the above Figure 1 Step 101 in the embodiment is similar, and the repeated parts are not repeated here.

[0070] It should be noted that, since the target decoration renderings finally generated in this embodiment are generated by an artificial intelligence model, the target decoration renderings only reflect the target decoration renderings for reference under specific constraints of the functional areas of the apartment represented by the target rough house picture, and there is no correlation between the target decoration renderings corresponding to the same target rough house picture and the multiple independently generated target decoration renderings, that is, each generated target decoration rendering is independent and for reference only. That is, the target decoration renderings are only used to quickly understand the user's decoration style and the details that the user cares about in decoration, provide direction for formal modeling and rendering, and reduce useless work.

[0071] That is to say, the decoration renderings generated corresponding to the rough house pictures of the functional areas of the same room taken from different shooting angles may be different, and even the target decoration effects generated multiple times from the same target rough house picture may be different. In view of this, in order to avoid users mistakenly believing that the rough house decoration effect generation method based on artificial intelligence in this embodiment can generate decoration renderings that correspond to all details of a set or the entire apartment, this step can be set to obtain only one target rough house picture each time.

[0072] 202. Use the first preset artificial intelligence model to perform deep processing on the target rough house image to obtain a deep assessment image of the target rough house.

[0073] The execution of this step is the same as the above Figure 1The step 102 in the embodiment is similar, and the repeated parts are not described here.

[0074] 203. Segment the target raw house picture using a second preset artificial intelligence model to obtain a target raw house segmentation map.

[0075] The execution of this step is similar to the foregoing Figure 1 The step 103 in the embodiment is similar, and the repeated parts are not described here.

[0076] 204. Input the target raw house depth evaluation map and the target raw house segmentation map into a third preset artificial intelligence model to output picture constraints of the target raw house picture.

[0077] The execution of this step is similar to the foregoing Figure 1 The step 104 in the embodiment is similar, and the repeated parts are not described here.

[0078] 205. Input the target raw house picture, the picture constraint information, and a preset creation prompt word into a fourth preset artificial intelligence model to obtain a target decoration effect picture corresponding to the target raw house picture.

[0079] Specifically, the target raw house picture obtained in step 201, the picture constraints obtained in step 204, and a preset creation prompt word directly input or selected by the user are input into the fourth preset artificial intelligence model to obtain a target decoration effect picture corresponding to the target raw house picture. For example, the preset creation prompt word is that the space type is selected as a living room, and the style type is selected as a light luxury style, etc., for example Figure 8 as shown; and finally generate Figure 8 the target decoration effect picture shown on the right, that is Figure 9 . Further, the fourth preset artificial intelligence model in this embodiment is a stablediffusion model. For details, please refer to Figure 11 and Figure 12 , Figure 11 as shown. Figure 6 Figure 7 as shown. Figure 9

[0080] ​​It should be noted that Stable Diffusion is a deep learning image generation model that generates images based on input prompts. ControlNet is an extended model that, when used in conjunction with Stable Diffusion, can control the structure of the output image. For example, using ControlNet's line segment detection model, the output image can be drawn based on line segments, for example Figure 10 As shown. The ControlNet model has many categories, including line segment, edge detection, depth, human posture, semantic segmentation, and so on. This example uses the Stable Diffusion deep learning model and the ControlNet model to achieve the function of directly outputting the decoration renderings of a rough house.

[0081] In this embodiment, training data collection and model training for the depth assessment model and segmentation model are performed. The depth model is responsible for perceiving the depth information of the content of the unfinished house image (evaluating the near-far relationship within the 3D scene); the segmentation model is responsible for perceiving the object category information (such as the floor, windows, ceiling, walls, etc.) of the content of the unfinished house image. During the model training process, the innovative use of unfinished house -> renovated house image pair data is made, fully leveraging the model generalization ability of deep learning to achieve the generation of unfinished house -> renovated house image.

[0082] The above embodiment describes the embodiment of the method for generating the decoration effect of rough house based on artificial intelligence of the present application. The following describes the device for generating the decoration effect of rough house based on artificial intelligence of the present application. Figure 3 An embodiment of the device for generating the decoration effect of a rough house based on artificial intelligence in this application includes:

[0083] An acquisition unit 301 is used to acquire a target rough housing picture;

[0084] Processing unit 302 is configured to perform depth processing on the target rough housing image using a first preset artificial intelligence model to obtain a depth assessment image of the target rough housing, wherein the first preset artificial intelligence model is a trained model capable of identifying the depth of each region of the rough housing image;

[0085] a segmentation unit 303 for segmenting the target rough housing image using a second preset artificial intelligence model to obtain a target rough housing segmentation map, wherein the second preset artificial intelligence model is a trained model capable of recognizing various building structure types in the rough housing image;

[0086] The constraint unit 304 is configured to input the target rough house depth evaluation graph and the target rough house segmentation graph into a third preset artificial intelligence model, and output picture constraints of the target rough house picture, the third preset artificial intelligence model being a trained model capable of generating picture constraints corresponding to a rough house picture according to a rough house depth evaluation graph and a rough house segmentation graph corresponding to the rough house picture.

[0087] The generation unit 305 is configured to generate a target decoration effect picture according to the target rough house picture, the picture constraints and input preset creative prompt words.

[0088] Optionally, when the generation unit 305 generates the target decoration effect picture according to the target rough house picture, the picture constraint information and input preset creative prompt words, the generation unit 305 is specifically configured to:

[0089] input the target rough house picture, the picture constraint information and the preset creative prompt words into a fourth preset artificial intelligence model to obtain a target decoration effect picture corresponding to the target rough house picture, the fourth preset artificial intelligence model being a trained model capable of generating a decoration effect picture corresponding to a rough house picture according to the rough house picture, picture constraint information corresponding to the rough house picture and input creative prompt words.

[0090] Optionally, the first preset artificial intelligence model is a depth_midas depth evaluation model.

[0091] Optionally, the second preset artificial intelligence model is a seg_ofade20k segmentation model.

[0092] Optionally, the third preset artificial intelligence model is a controlnet model.

[0093] Optionally, the fourth preset artificial intelligence model is a stablediffusion model.

[0094] The operations performed by the rough house decoration effect generation device based on artificial intelligence in the present application are similar to the operations described in the foregoing Figure 1 or Figure 2 embodiments, and thus will not be described here in detail.

[0095] The computer device in the embodiments of the present application will be described below. Please refer to Figure 4 One embodiment of the computer device in the embodiments of the present application includes:

[0096] The computer device 400 can include one or more processors (central processing units, CPUs) 401 and a memory 402 in which one or more applications or data are stored. The memory 402 is volatile storage or persistent storage. The programs stored in the memory 402 can include one or more modules, each of which can include a series of instruction operations in the computer device. Further, the processor 401 can be configured to communicate with the memory 402 to execute the series of instruction operations in the memory 402 on the computer device 400. The computer device 400 can also include one or more wireless network interfaces 403, one or more input / output interfaces 404, and / or one or more operating systems, such as HarmonyOS, Windows Server, Mac OS, Unix, Linux, FreeBSD, etc. The processor 401 can execute the operations performed in any of the embodiments described above, and details are not repeated here. Figure 1 or Figure 2 The operations performed in any of the embodiments described above can be executed by the processor 401, and details are not repeated here.

[0097] In several embodiments provided in the present application, those skilled in the art should understand that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and in actual implementation, another division mode can be used, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0098] In addition, the functional units in the various embodiments of the present application can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit. If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present application. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0099] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements or improvements made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A method for generating decoration effects of rough houses based on artificial intelligence, characterized in that: include: Get the target rough house picture; Performing depth processing on the target rough housing image using a first preset artificial intelligence model to obtain a depth assessment map of the target rough housing, wherein the first preset artificial intelligence model is a trained model capable of identifying the depth of each area of ​​the rough housing image; Segmenting the target rough housing image using a second preset artificial intelligence model to obtain a target rough housing segmentation map, wherein the second preset artificial intelligence model is a trained model capable of identifying various building structure types in the rough housing image; Inputting the target rough house depth assessment map and the target rough house segmentation map into a third preset artificial intelligence model, and outputting the image constraints of the target rough house picture, wherein the third preset artificial intelligence model is a trained model capable of generating the image constraints corresponding to the rough house picture based on the rough house depth assessment map and the rough house segmentation map corresponding to the rough house picture; A target decoration rendering is generated based on the target rough house picture, the picture constraints and the input preset creation prompt words.

2. The method for generating decoration effects of rough houses based on artificial intelligence according to claim 1 is characterized in that: Generating a target decoration rendering according to the target rough house picture, the picture constraint information, and the input preset creation prompt words includes: The target unfinished house picture, the picture constraint information and the preset creation prompt words are input into the fourth preset artificial intelligence model to obtain the target decoration rendering corresponding to the target unfinished house picture. The fourth preset artificial intelligence model is a trained model that can generate the decoration rendering corresponding to the unfinished house picture based on the unfinished house picture, the picture constraint information corresponding to the unfinished house picture and the input creation prompt words.

3. The method for generating decoration effects of rough houses based on artificial intelligence according to claim 1 is characterized in that: The first preset artificial intelligence model is the depth_midas depth assessment model.

4. The method for generating decoration effects of rough housing based on artificial intelligence according to claim 1 is characterized in that: The second preset artificial intelligence model is the seg_ofade20k segmentation model.

5. The method for generating decoration effects of rough housing based on artificial intelligence according to claim 1 is characterized in that: The third preset artificial intelligence model is a controlnet model.

6. The method for generating decoration effects of rough housing based on artificial intelligence according to claim 2 is characterized in that: The fourth preset artificial intelligence model is a stable diffusion model.

7. A device for generating decoration effects of rough houses based on artificial intelligence, characterized in that: include: An acquisition unit, used to acquire a target rough housing picture; a processing unit, configured to perform depth processing on the target rough housing image using a first preset artificial intelligence model to obtain a depth assessment image of the target rough housing, wherein the first preset artificial intelligence model is a trained model capable of identifying the depth of each region of the rough housing image; a segmentation unit, configured to segment the target rough housing image using a second preset artificial intelligence model to obtain a target rough housing segmentation map, wherein the second preset artificial intelligence model is a trained model capable of recognizing various building structure types in the rough housing image; a constraint unit, configured to input the target rough house depth assessment map and the target rough house segmentation map into a third preset artificial intelligence model, and output image constraints for the target rough house image, wherein the third preset artificial intelligence model is a trained model capable of generating image constraints corresponding to the rough house image based on the rough house depth assessment map and the rough house segmentation map corresponding to the rough house image; A generating unit is used to generate a target decoration rendering based on the target rough house picture, the picture constraints and the input preset creation prompt words.

8. A computer device, characterized in that: include: Processor, memory, bus, input and output interface, network interface; The processor is connected to the memory, the input / output interface, and the network interface via a bus; The memory stores a program; When the processor executes the program stored in the memory, it implements the method for generating the decoration effect of a rough house based on artificial intelligence as described in any one of claims 1 to 6.

9. A computer storage medium, characterized in that The computer storage medium stores instructions, and when the instructions are executed on the computer, the computer executes the method for generating the decoration effect of a rough house based on artificial intelligence as described in any one of claims 1 to 6.

10. A computer program product, characterized in that When the computer program product is executed on a computer, the computer is caused to execute the method for generating the decoration effect of a rough house based on artificial intelligence as described in any one of claims 1 to 6.

Images