CN115222836A - Image generation method, device and equipment based on counterstudy and storage medium - Google Patents
Image generation method, device and equipment based on counterstudy and storage medium Download PDFInfo
- Publication number
- CN115222836A CN115222836A CN202210672994.6A CN202210672994A CN115222836A CN 115222836 A CN115222836 A CN 115222836A CN 202210672994 A CN202210672994 A CN 202210672994A CN 115222836 A CN115222836 A CN 115222836A
- Authority
- CN
- China
- Prior art keywords
- image
- motorcycle
- target
- image generation
- similar
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Biophysics (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The application relates to the technical field of image information processing, and provides an image generation method, device, equipment and storage medium based on counterstudy, wherein the method is based on a preset motorcycle image data set and a counterstudy algorithm, and is used for training a preset model to generate a motorcycle image generation model; and generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model. Through the mode, the target motorcycle image input by the user is converted into the similar motorcycle image based on the motorcycle image generation model, so that the similar motorcycle image fused with the original motorcycle image characteristic can be automatically generated, and the problems of low image generation efficiency and poor user experience are solved.
Description
Technical Field
The present invention relates to the field of image information processing technologies, and in particular, to an image generation method, an image generation device, an image generation apparatus, and a computer-readable storage medium based on counterstudy.
Background
With the rapid development of intelligent terminal equipment, the hardware performance and the operation function of the intelligent terminal equipment are improved, the image quality of pictures shot by a camera is better and better, and a plurality of applications of the intelligent terminal equipment combined with the operation function appear, such as: the intelligent image information processing technology can be used for designing and manufacturing machines due to the functions of image editing, background replacement, image synthesis and the like.
In the prior art, because the mechanical design modeling workload is huge, when a designer wants to iterate a design drawing according to an intelligent image, a method for automatically generating a pixel picture is not always available, and the image generation efficiency is low; in addition, the existing automatic generation method of the motorcycle design drawing has the conditions of rough image details, incomplete outline and unreality of texture generated for motorcycle style, motorcycle texture and motorcycle appearance, and the user experience is poor.
Disclosure of Invention
Based on the above, the application provides an image generation method based on counterstudy, and aims to solve the problems of low image generation efficiency and poor user experience in the prior art.
To achieve the above object, the present invention provides an image generation method based on counterstudy, including: training a preset model based on a preset motorcycle image data set and a counterstudy learning algorithm to generate a motorcycle image generation model; and generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model.
Further, the generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model includes:
performing feature extraction on the target motorcycle image through the motorcycle image generation model to obtain target image features of the target motorcycle image, wherein the target image features comprise target style features, target texture features and target appearance features;
and decoding the target style characteristic, the target texture characteristic and the target appearance characteristic through the motorcycle image generation model to generate an image similar to the style, texture and appearance of the motorcycle in the target motorcycle image as the similar motorcycle image.
Further, the generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model includes:
based on the motorcycle image generation model, carrying out style feature extraction on the target motorcycle image to generate target style image features;
and decoding the target style image characteristics through the motorcycle image generation model to generate an image with a style similar to that of the motorcycle in the target motorcycle image as the similar motorcycle image.
Further, the generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model includes:
based on the motorcycle image generation model, performing texture feature extraction on the target motorcycle image to generate target texture image features;
and decoding the target texture image features through the motorcycle image generation model to generate an image similar to the texture of the motorcycle in the target motorcycle image as the similar motorcycle image.
Further, the generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model includes:
based on the motorcycle image generation model, extracting appearance characteristics of the target motorcycle image to generate target appearance image characteristics;
and decoding the target appearance image characteristics through the motorcycle image generation model to generate an image similar to the appearance of the motorcycle in the target motorcycle image as the similar motorcycle image.
Further, the motorcycle image generation model comprises a generator, a discriminator and a feature extractor; the training of the preset model based on the preset motorcycle image data set and the counterstudy algorithm to generate the motorcycle image generation model comprises the following steps:
collecting different motorcycle images as the motorcycle image dataset;
randomly sampling a motorcycle image from the motorcycle image data set as a training motorcycle image, extracting a first image feature vector of the training motorcycle image based on the feature extractor, and outputting the first image feature vector of the training motorcycle image;
inputting the first image feature vector into the generator, and calculating and outputting a predicted motorcycle image through each layer of neurons in the generator;
extracting a second image feature vector of the predicted motorcycle image through the feature extractor, and outputting the second image feature vector;
inputting a preset standard image feature vector and the second image feature vector into the discriminator as a pair of feature vectors, and outputting the probability that the second image feature vector is identified as the preset standard image feature vector;
and when the probability is not less than a preset value, finishing the training of the preset model and generating a motorcycle image generation model.
Further, after the inputting a preset standard image feature vector and the second image feature vector as a pair of feature vectors into the discriminator and outputting the probability that the second feature vector is recognized as the preset standard image feature vector, the method further includes:
and when the probability is smaller than a preset value, setting a cycle coordination loss function, continuously iterating the loss of the characteristic vectors of the two parties in a cycle calculation mode, calculating the network weight gradient by adopting back propagation and updating the parameters of the image generation model for counterstudy, and stopping training the motorcycle image generation model until the probability is not smaller than the preset value to generate the motorcycle image generation model.
Further, to achieve the above object, the present invention also provides an image generation apparatus based on counterstudy, comprising: the image model generating module is used for training a preset model based on a preset motorcycle image data set and a counterstudy algorithm to generate a motorcycle image generating model; and the similar image generation module is used for generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model.
Furthermore, in order to achieve the above object, the present invention also provides an image generation apparatus based on resist learning, which includes a processor, a memory, and an image generation program based on resist learning stored on the memory and executable by the processor, wherein when the image generation program based on resist learning is executed by the processor, the steps of the image generation method based on resist learning as described above are implemented.
Further, to achieve the above object, the present invention also provides a computer-readable storage medium having stored thereon a counterlearning-based image generation program, wherein when the counterlearning-based image generation program is executed by a processor, the steps of the counterlearning-based image generation method as described above are implemented.
The invention provides an image generation method based on counterstudy, which is characterized in that a preset model is trained based on a preset motorcycle image data set and a counterstudy algorithm to generate a motorcycle image generation model; and generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model. Through the mode, the target motorcycle image input by the user is converted into the similar motorcycle image based on the motorcycle image generation model, so that the similar motorcycle image fused with the original motorcycle image characteristic can be automatically generated, and the problems of low image generation efficiency and poor user experience are solved.
Drawings
Fig. 1 is a schematic diagram of a hardware configuration of an image generation apparatus based on countermeasure learning according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating a first embodiment of an image generation method based on counterlearning according to the present invention;
FIG. 3 is a flowchart illustrating a second embodiment of an image generation method based on counterlearning according to the present invention;
FIG. 4 is a flowchart illustrating a third embodiment of an image generation method based on counterlearning according to the present invention;
fig. 5 is a functional block diagram of the image generating apparatus based on the counterstudy according to the first embodiment of the present invention.
The implementation, functional features and advantages of the present invention will be further described with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The image generation method based on the counterstudy related to the embodiment of the invention is mainly applied to the image generation device based on the counterstudy, and the image generation device based on the counterstudy can be a device with display and processing functions, such as a PC, a portable computer, a mobile terminal and the like.
Referring to fig. 1, fig. 1 is a schematic diagram of a hardware configuration of an image generation apparatus based on countermeasure learning according to an embodiment of the present invention.
In an embodiment of the present invention, the image generation apparatus based on the counterlearning may include a processor 1001 (e.g., CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used for realizing connection communication among the components; the user interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard); the network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface); the memory 1005 may be a high-speed RAM memory, or may be a non-volatile memory (e.g., a magnetic disk memory), and optionally, the memory 1005 may be a storage device independent of the processor 1001.
Those skilled in the art will appreciate that the hardware configuration shown in fig. 1 does not constitute a limitation of the resist learning-based image generation apparatus, and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
With continued reference to fig. 1, the memory 1005 of fig. 1, which is one type of computer-readable storage medium, may include an operating system, a network communication module, and an image generation program based on counterlearning.
In fig. 1, the network communication module is mainly used for connecting to a server and performing data communication with the server; and the processor 1001 may call the counterlearning-based image generation program stored in the memory 1005 and execute the counterlearning-based image generation method provided by the embodiment of the present invention.
The embodiment of the invention provides an image generation method based on counterstudy.
Referring to fig. 2, fig. 2 is a schematic flow chart of a first embodiment of the image generation method based on the counterstudy according to the present invention.
In this embodiment, the image generation method based on the counterstudy includes the steps of:
s10, training a preset model based on a preset motorcycle image data set and a counterstudy learning algorithm to generate a motorcycle image generation model;
in this embodiment, a large number of different motorcycle images are collected first, a preset motorcycle image data set is constructed, then the preset motorcycle image data set and a counterstudy learning algorithm are used for training a motorcycle image generation model, and finally the generator is reserved as a final motorcycle image generation model.
Specifically, a large number of motorcycle images of different styles, textures or appearances are collected in advance as a motorcycle image data set, wherein the number of images in the motorcycle data set is at least 1, and one of the motorcycle images is randomly extracted as a training motorcycle image.
Further, the motorcycle image generation model comprises a generator, a discriminator and a feature extractor; the method for training the preset model based on the preset motorcycle image data set and the confrontation learning algorithm to generate the motorcycle image generation model comprises the following steps:
collecting different motorcycle images as the motorcycle image dataset;
randomly sampling a motorcycle image from the motorcycle image data set as a training motorcycle image, extracting a first image feature vector of the training motorcycle image based on the feature extractor, and outputting the first image feature vector of the training motorcycle image;
inputting the first image feature vector into the generator, and calculating and outputting a predicted motorcycle image through each layer of neurons in the generator;
extracting a second image feature vector of the predicted motorcycle image through the feature extractor, and outputting the second image feature vector;
inputting a preset standard image feature vector and the second image feature vector into the discriminator as a pair of feature vectors, and outputting the probability that the second image feature vector is identified as the preset standard image feature vector;
and when the probability is not less than a preset value, finishing the training of the preset model and generating a motorcycle image generation model.
In this embodiment, an image generation model based on counterstudy is trained; the method comprises the steps of inputting at least 1 training motorcycle image into a generator, obtaining a prediction motorcycle image converted by the training motorcycle image, respectively inputting the prediction motorcycle image and a corresponding standard motorcycle image (namely the input training motorcycle image) into a feature extractor and a discriminator, outputting style, texture and appearance feature vectors of the prediction motorcycle image and the corresponding standard motorcycle image and the probability of the style, the texture and the appearance feature vectors of the prediction motorcycle image and the standard motorcycle image recognized by the discriminator, calculating and updating the weight gradient of a neural network of an anti-learning image model in a back propagation mode, and repeatedly performing the process until the learning effect of the generator on the style, the texture and the appearance feature of the motorcycle image meets the requirement.
It can be understood that, in this embodiment, the generator uses a neural network model, which is equivalent to a nonlinear function, and can implement mapping from input to output, the feature vector of the input image is input to the network, and will perform nonlinear calculation with the neurons in the network, the neurons in the lower layer of the network are combined with each other with a certain weight, and then output to the next layer of neurons, and finally output the desired result at the output layer, the training motorcycle image is input to the generator, and the prediction motorcycle image is output via the nonlinear combination of the neurons in each layer of the generator network;
in this embodiment, the arbiter also employs a neural network, which has the advantages that only the training data and the standard image need to be prepared, the network training can automatically learn according to the standard image, the neural network parameters are continuously updated iteratively through the forward and backward propagation of the neural network in the middle, and the desired result is obtained after the training is finished. The arbiter learns the standard image and the predicted image at the same time, and distinguishes them as much as possible, the output is a value of 0-1, which represents the probability that the image input to the arbiter is the standard image, the output probability value and the characteristic matrix of the standard image and the predicted image respectively participate in the loss calculation of the cyclic loss function, the loss function updates the parameter of each neuron in the neural network through the back propagation process, namely the training process of the neural network;
in this embodiment, the feature extractor is capable of extracting features of the input standard image and the corresponding predicted image, including style, texture and appearance features of the motorcycle image.
And S20, generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model.
In the embodiment, the user can shoot the motorcycle seen on the street or store any target motorcycle image seen in the computer, and can output the similar motorcycle image with the similar characteristics to the target motorcycle image by inputting any target motorcycle image to the image generation model, and the motorcycle design drawing with the same characteristics as the target motorcycle image can be obtained without redundant input and operation of the user.
Referring to fig. 3, fig. 3 is a flowchart illustrating a second embodiment of the image generation method based on the counterstudy according to the present invention.
Based on the foregoing embodiment shown in fig. 2, in this embodiment, the step S20 specifically includes:
step S21, performing feature extraction on the target motorcycle image through the motorcycle image generation model to obtain target image features of the target motorcycle image, wherein the target image features comprise target style features, target texture features and target appearance features;
and S22, decoding the target style characteristic, the target texture characteristic and the target appearance characteristic through the motorcycle image generation model, and generating an image similar to the style, texture and appearance of the motorcycle in the target motorcycle image as the similar motorcycle image.
In this embodiment, it is assumed that a device is equipped with the image generation system based on the counterlearning, and a user inputs a "target motorcycle image 1", the image generation system based on the counterlearning provides an option of outputting a motorcycle image having similar style, texture and appearance characteristics to the target image 1, and the motorcycle image generation model generates the "similar motorcycle image 1" having similar style, texture and appearance characteristics to the target image 1 when the user selects the option.
Referring to fig. 4, fig. 4 is a flowchart illustrating a third embodiment of the image generation method based on the counterstudy according to the present invention.
Based on the foregoing embodiment shown in fig. 2, in this embodiment, the step S20 further includes:
step S23, based on the motorcycle image generation model, carrying out style feature extraction on the target motorcycle image to generate target style image features;
and S24, decoding the target style image characteristics through the motorcycle image generation model, and generating an image with a style similar to that of a motorcycle in the target motorcycle image as the similar motorcycle image.
It can be understood that, in this embodiment, texture feature extraction may be performed on the target motorcycle image based on the motorcycle image generation model to generate a target texture image feature; and decoding the target texture image features through the motorcycle image generation model to generate an image similar to the texture of the motorcycle in the target motorcycle image as the similar motorcycle image.
It can be understood that, in this embodiment, texture feature extraction may be performed on the target motorcycle image based on the motorcycle image generation model to generate a target texture image feature; and decoding the target texture image features through the motorcycle image generation model to generate an image similar to the texture of the motorcycle in the target motorcycle image as the similar motorcycle image.
In this embodiment, it is assumed that a device is equipped with the image generation system based on the countermeasure learning, and a user inputs a "target motorcycle image 1'", and the image generation system based on the countermeasure learning provides three options, that is, three motorcycle images P1', P2', and P3' having similar styles, textures, or appearance characteristics to the target image 1 'are output, respectively, and when the user selects this option, the motorcycle image generation model can generate three pictures having similar styles to the target image 1', that is, three pictures of a "similar motorcycle image P1'", "similar motorcycle image P2'", and "similar motorcycle image P3", respectively.
In addition, the embodiment of the invention also provides an image generation device based on counterstudy.
Referring to fig. 5, fig. 5 is a functional block diagram of a first embodiment of an image generating apparatus based on counterlearning according to the present invention.
In this embodiment, the image generation device based on the counterstudy includes:
the image model generation module 10 is used for training a preset model based on a preset motorcycle image data set and a counterstudy learning algorithm to generate a motorcycle image generation model;
and the similar image generation module 20 is used for generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model.
Further, the similar image generating module 20 specifically includes:
the image feature extraction unit is used for performing feature extraction on the target motorcycle image through the motorcycle image generation model to obtain target image features of the target motorcycle image, wherein the target image features comprise target style features, target texture features and target appearance features;
and the similar image generation unit decodes the target style characteristic, the target texture characteristic and the target appearance characteristic through the motorcycle image generation model to generate an image similar to the style, texture and appearance of the motorcycle in the target motorcycle image as the similar motorcycle image.
Further, the similar image generating module 20 specifically includes:
the style feature extraction unit is used for extracting style features of the target motorcycle image based on the motorcycle image generation model to generate target style image features;
and a similar style image generation unit which decodes the target style image characteristics through the motorcycle image generation model and generates an image similar to the style of the motorcycle in the target motorcycle image as the similar motorcycle image.
Further, the image model generating module 20 specifically further includes:
the texture feature extraction unit is used for extracting texture features of the target motorcycle image based on the motorcycle image generation model to generate target texture image features;
and a similar texture image generation unit which decodes the target texture image features through the motorcycle image generation model and generates an image similar to the texture of the motorcycle in the target motorcycle image as the similar motorcycle image.
Further, the image model generating module 20 specifically further includes:
an appearance image generation unit which extracts appearance characteristics of the target motorcycle image based on the motorcycle image generation model to generate target appearance image characteristics;
and a similar appearance image generation unit which decodes the target appearance image features through the motorcycle image generation model and generates an image similar to the appearance of the motorcycle in the target motorcycle image as the similar motorcycle image.
Further, the image model generation module 10 includes:
an image dataset unit that collects different motorcycle images as the motorcycle image dataset;
a feature vector extraction unit which randomly samples a motorcycle image from the motorcycle image data set as a training motorcycle image, extracts a first image feature vector of the training motorcycle image based on the feature extractor, and outputs the first image feature vector of the training motorcycle image;
a prediction image generation unit which inputs the first image feature vector into the generator and outputs a prediction motorcycle image through calculation of neurons in each layer in the generator;
a second feature vector extraction unit that extracts a second image feature vector of the predicted motorcycle image by the feature extractor and outputs the second image feature vector;
a discriminator determination unit that inputs a preset standard image feature vector and the second image feature vector as a pair of feature vectors to the discriminator and outputs a probability that the second image feature vector is recognized as the preset standard image feature vector;
and the model generation unit is used for finishing the training of the preset model and generating a motorcycle image generation model when the probability is not less than the preset value.
Further, the image model generation module 10 further includes:
and the function circulation unit is used for setting a circulation coordination loss function when the probability is smaller than a preset value, continuously iterating the loss of the characteristic vectors of the two parties in a circulation calculation mode, calculating a network weight gradient by adopting back propagation and updating the image generation model parameters of the counterstudy, and stopping training the motorcycle image generation model until the probability is not smaller than the preset value so as to generate the motorcycle image generation model.
Each module in the image generation device based on the counterstudy corresponds to each step in the embodiment of the image generation method based on the counterstudy, and the functions and the implementation process of the image generation device are not described in detail herein.
In addition, the embodiment of the invention also provides a computer readable storage medium.
The computer-readable storage medium of the present invention stores thereon a counterlearning-based image generation program, wherein the counterlearning-based image generation program, when executed by a processor, implements the steps of the counterlearning-based image generation method as described above.
The method implemented when the counterlearning-based image generation program is executed may refer to various embodiments of the counterlearning-based image generation method of the present invention, and details thereof are not repeated herein.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising a … …" does not exclude the presence of another identical element in a process, method, article, or system that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. An image generation method based on counterlearning, characterized by comprising the steps of:
training a preset model based on a preset motorcycle image data set and a confrontation learning algorithm to generate a motorcycle image generation model;
and generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model.
2. The method for generating an image based on antagonistic learning according to claim 1, wherein said generating a similar motorcycle image corresponding to a target motorcycle image by said motorcycle image generation model comprises:
performing feature extraction on the target motorcycle image through the motorcycle image generation model to obtain target image features of the target motorcycle image, wherein the target image features comprise target style features, target texture features and target appearance features;
and decoding the target style characteristic, the target texture characteristic and the target appearance characteristic through the motorcycle image generation model to generate an image similar to the style, texture and appearance of the motorcycle in the target motorcycle image as the similar motorcycle image.
3. The method for generating an image based on antagonistic learning according to claim 1, wherein said generating a similar motorcycle image corresponding to a target motorcycle image by said motorcycle image generation model comprises:
based on the motorcycle image generation model, carrying out style feature extraction on the target motorcycle image to generate target style image features;
and decoding the target style image characteristics through the motorcycle image generation model to generate an image similar to the style of the motorcycle in the target motorcycle image as the similar motorcycle image.
4. The method for generating an image based on antagonistic learning according to claim 1, wherein said generating a similar motorcycle image corresponding to a target motorcycle image by said motorcycle image generation model comprises:
based on the motorcycle image generation model, extracting texture features of the target motorcycle image to generate target texture image features;
and decoding the target texture image features through the motorcycle image generation model to generate an image similar to the texture of the motorcycle in the target motorcycle image as the similar motorcycle image.
5. The method for generating an image based on antagonistic learning according to claim 1, wherein said generating a similar motorcycle image corresponding to a target motorcycle image by said motorcycle image generation model comprises:
based on the motorcycle image generation model, extracting appearance characteristics of the target motorcycle image to generate target appearance image characteristics;
and decoding the target appearance image characteristics through the motorcycle image generation model to generate an image similar to the appearance of the motorcycle in the target motorcycle image as the similar motorcycle image.
6. The method of image generation based on antagonistic learning according to any one of claims 1 to 5, characterised in that the motorcycle image generation model comprises a generator, a discriminator and a feature extractor; the training of the preset model based on the preset motorcycle image data set and the counterstudy algorithm to generate the motorcycle image generation model comprises the following steps:
collecting different motorcycle images as the motorcycle image dataset;
randomly sampling a motorcycle image from the motorcycle image data set as a training motorcycle image, extracting a first image feature vector of the training motorcycle image based on the feature extractor, and outputting the first image feature vector of the training motorcycle image;
inputting the first image feature vector into the generator, and calculating and outputting a predicted motorcycle image through each layer of neurons in the generator;
extracting a second image feature vector of the predicted motorcycle image through the feature extractor, and outputting the second image feature vector;
inputting a preset standard image feature vector and the second image feature vector into the discriminator as a pair of feature vectors, and outputting the probability that the second image feature vector is identified as the preset standard image feature vector;
and when the probability is not less than a preset value, finishing the training of the preset model and generating a motorcycle image generation model.
7. The method of generating an image based on counterlearning according to claim 6, wherein after inputting a preset standard image feature vector and the second image feature vector as a pair of feature vectors into the discriminator and outputting a probability that the second feature vector is recognized as the preset standard image feature vector, the method further comprises:
and when the probability is smaller than a preset value, setting a cycle coordination loss function, continuously iterating the loss of the characteristic vectors of the two parties in a cycle calculation mode, calculating the network weight gradient by adopting back propagation and updating the parameters of the image generation model for counterstudy, and stopping training the motorcycle image generation model until the probability is not smaller than the preset value to generate the motorcycle image generation model.
8. An image generation apparatus based on antagonistic learning, characterized by comprising:
the image model generating module is used for training a preset model based on a preset motorcycle image data set and a counterstudy algorithm to generate a motorcycle image generating model;
and the similar image generation module is used for generating a similar motorcycle image corresponding to the target motorcycle image through the motorcycle image generation model.
9. An image generation apparatus based on resist learning, characterized in that the apparatus comprises a processor, a memory, and an image generation program based on resist learning stored on the memory and executable by the processor, wherein the image generation program based on resist learning realizes the steps of the image generation method based on resist learning according to any one of claims 1 to 7 when executed by the processor.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a counterlearning-based image generation program, wherein the counterlearning-based image generation program, when executed by a processor, implements the steps of the counterlearning-based image generation method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210672994.6A CN115222836A (en) | 2022-06-15 | 2022-06-15 | Image generation method, device and equipment based on counterstudy and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210672994.6A CN115222836A (en) | 2022-06-15 | 2022-06-15 | Image generation method, device and equipment based on counterstudy and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115222836A true CN115222836A (en) | 2022-10-21 |
Family
ID=83607089
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210672994.6A Pending CN115222836A (en) | 2022-06-15 | 2022-06-15 | Image generation method, device and equipment based on counterstudy and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115222836A (en) |
-
2022
- 2022-06-15 CN CN202210672994.6A patent/CN115222836A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xiong et al. | Learning to generate time-lapse videos using multi-stage dynamic generative adversarial networks | |
| CN112418292B (en) | Image quality evaluation method, device, computer equipment and storage medium | |
| CN109685121B (en) | Training method of image retrieval model, image retrieval method and computer equipment | |
| CN111476708B (en) | Model generation method, model acquisition method, device, equipment and storage medium | |
| CN110322416A (en) | Image processing method, device and computer readable storage medium | |
| CN112990390B (en) | Training method of image recognition model, and image recognition method and device | |
| WO2015062209A1 (en) | Visualized optimization processing method and device for random forest classification model | |
| CN109472764B (en) | Method, apparatus, device and medium for image synthesis and image synthesis model training | |
| US20220284678A1 (en) | Method and apparatus for processing face information and electronic device and storage medium | |
| CN112381707B (en) | Image generation method, device, equipment and storage medium | |
| CN113822790B (en) | Image processing method, device, equipment and computer readable storage medium | |
| CN112950640A (en) | Video portrait segmentation method and device, electronic equipment and storage medium | |
| CN111539353A (en) | Image scene recognition method and device, computer equipment and storage medium | |
| JP2023545052A (en) | Image processing model training method and device, image processing method and device, electronic equipment, and computer program | |
| CN114202615A (en) | Facial expression reconstruction method, device, equipment and storage medium | |
| CN114170484B (en) | Picture attribute prediction method and device, electronic equipment and storage medium | |
| CN115984447A (en) | Image rendering method, device, equipment and medium | |
| CN113628101B (en) | Three-stage tile image generation method based on GAN network structure | |
| WO2024087946A1 (en) | Image editing method and apparatus, computer device, and storage medium | |
| CN111914106B (en) | Texture and normal library construction method, texture and normal map generation method and device | |
| CN117237547A (en) | Image reconstruction method, reconstruction model processing method and device | |
| CN117557708A (en) | Image generation method, device, storage medium and computer equipment | |
| CN115222836A (en) | Image generation method, device and equipment based on counterstudy and storage medium | |
| CN115601479A (en) | Virtual human head generation method and device | |
| CN114373033A (en) | Image processing method, image processing apparatus, image processing device, storage medium, and computer program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |