CN116452906B - Railway wagon fault picture generation method based on text description - Google Patents
Railway wagon fault picture generation method based on text description Download PDFInfo
- Publication number
- CN116452906B CN116452906B CN202310197794.4A CN202310197794A CN116452906B CN 116452906 B CN116452906 B CN 116452906B CN 202310197794 A CN202310197794 A CN 202310197794A CN 116452906 B CN116452906 B CN 116452906B
- Authority
- CN
- China
- Prior art keywords
- fault
- image
- text description
- generating
- railway wagon
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000007781 pre-processing Methods 0.000 claims abstract description 23
- 238000012549 training Methods 0.000 claims abstract description 17
- 238000001514 detection method Methods 0.000 claims abstract description 15
- 239000013598 vector Substances 0.000 claims description 34
- 238000012545 processing Methods 0.000 claims description 21
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000012216 screening Methods 0.000 claims description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 238000013136 deep learning model Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Classifications
-
- 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
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/10—Text processing
- G06F40/12—Use of codes for handling textual entities
- G06F40/126—Character encoding
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
- G06V10/16—Image acquisition using multiple overlapping images; Image stitching
-
- 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
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Image Analysis (AREA)
Abstract
A method for generating a railway wagon fault picture based on text description belongs to the technical field of fault picture generation. The method aims at the problems that the efficiency of obtaining the railway wagon fault image by adopting the manual PS mode is low, and the training requirement of a fault detection model cannot be met. Comprising the following steps: acquiring failure-free sub-images of all parts of the railway wagon and describing expected failure forms by using texts; the artificial PS obtains an expected failure image of an expected failure form; generating a generated fault image of an expected fault form based on the fault-free sub-image and the pre-processed fault image by adopting a generator of the GAN network; training the GAN network by combining the expected failure image to obtain a trained GAN network generator; performing fault form text description on the image to be processed, and preprocessing the fault form text description to obtain a corresponding preprocessed fault image; and then generating a fault image based on the input image to be processed and the corresponding preprocessing fault image by adopting the trained GAN network generator. The method is used for generating the railway wagon fault picture.
Description
Technical Field
The invention relates to a method for generating a railway wagon fault picture based on text description, and belongs to the technical field of fault picture generation.
Background
The fault detection of the railway wagon by using the deep learning method needs a large number of fault images as a training set of the deep learning model, and the detection precision of the model has an important relation with the quantity and quality of the fault images in the training set.
Sources of conventional failure images mainly include two approaches: one is to collect the real fault image, and the other is to simulate the fault form by means of manual PS to obtain the fault image. For some faults with low occurrence probability, it is difficult to collect enough real fault images for model training, so that the real faults need to be simulated by means of a manual PS to obtain the fault images. The manual PS fault image requires a lot of time, and usually every person can PS40 images every day, and the inefficiency of the fault image becomes a main factor affecting the development speed of the detection model.
Disclosure of Invention
Aiming at the problems that the efficiency of obtaining the railway wagon fault image by adopting the manual PS mode is low and the training requirement of a fault detection model cannot be met, the invention provides a railway wagon fault image generation method based on text description.
The invention relates to a method for generating a railway wagon fault picture based on text description, which comprises the following steps of,
step one: acquiring fault-free sub-images of all parts of the railway wagon;
step two: describing expected fault forms by text for each fault-free sub-image;
step three: performing image processing on the failure-free sub-images according to the text description to obtain expected failure images of expected failure forms;
step four: preprocessing the text description to obtain word vectors, encoding all the word vectors to form feature vectors, and obtaining a preprocessed fault image based on the feature vectors; generating a generated fault image of an expected fault form based on the fault-free sub-image and the pre-processed fault image by adopting a generator of the GAN network; then adopting a discriminator of the GAN network to discriminate the expected fault image and the generated fault image until the GAN network reaches a steady state, and obtaining a trained GAN network generator;
step five: performing fault form text description on the image to be processed, and preprocessing the fault form text description to obtain a corresponding preprocessed fault image; and then a trained GAN network generator is adopted to generate a fault image based on the input image to be processed and the corresponding pre-processing fault image, and the fault image is used as training data of the fault detection network.
According to the method for generating the railway wagon fault picture based on the text description, the pretreatment process in the step four is the same as that in the step five;
the word vector obtaining method in the preprocessing comprises the following steps:
word segmentation processing is carried out on the text description, and word2vec method is adopted to obtain word vectors of each word.
According to the method for generating the railway wagon fault picture based on the text description, the method for obtaining the preprocessing fault image in the preprocessing comprises the following steps:
extracting features of all word vectors by adopting an LSTM network, and encoding to form feature vectors;
taking the feature vector as primary data with the size of 1x1 and the channel number as feature vector dimensions, and carrying out convolution operation for 64 times to obtain secondary data with the size of 8x8 and the channel number of 64; and performing deconvolution operation on the two-stage data for a plurality of times to obtain a pre-processed fault image, wherein the size of the pre-processed fault image is the same as that of the non-fault sub-image.
According to the method for generating the railway wagon fault picture based on the text description, U-net is adopted as a generator in the GAN network.
According to the method for generating the railway wagon fault picture based on the text description, the input image of the U-net generator further comprises a random noise image.
According to the railway wagon fault picture generation method based on text description, a U-net generator splices a 3-channel fault-free sub-image, a 1-channel pre-processing fault image and a 1-channel random noise image to obtain a 5-channel primary image, and then the 5-channel primary image is used for generating a fault image with an expected fault form.
According to the method for generating the railway wagon fault picture based on the text description, the fault image generated by the image to be processed in the fifth step is screened and qualified according to the preset standard and then is used as training data of a fault detection network.
According to the method for generating the railway wagon fault picture based on text description, in the first step, the obtaining method of the fault-free sub-image of each part comprises the following steps: and acquiring the railway wagon image by adopting a linear array camera, obtaining the position information of each part of the railway wagon by a target detection method, and obtaining fault-free sub-images of each part according to the position information.
According to the method for generating the railway wagon fault picture based on the text description, each part has 1000 fault-free sub-images.
According to the method for generating the railway wagon fault picture based on the text description, text description contents in the second step comprise that a brake cylinder falls off and inclines by 20 degrees.
The invention has the beneficial effects that: the method of the invention generates the fault image through the neural network model, and can rapidly complete the simulation of partial faults, thereby accelerating the development speed of the fault detection model.
The method of the invention utilizes the generation model to generate the fault image, reduces the workload of the artificial PS image and can improve the generation efficiency of the fault image.
The method uses the real image and the fault description text as input, can modify the image on the basis of the real image, and controls the wanted fault form and attribute through the text description; for example, the degree of tilting is controlled by text when generating the tilting member. Through text description, the generated model is more focused on the part needing to be modified in the image, and the effect of generating the fault image is better.
Drawings
Fig. 1 is a flowchart of a railway wagon fault picture generation method based on text description.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
It should be noted that, without conflict, the embodiments of the present invention and features of the embodiments may be combined with each other.
The invention is further described below with reference to the drawings and specific examples, which are not intended to be limiting.
The invention provides a method for generating a railway wagon fault picture based on text description, which is shown in the accompanying figure 1,
step one: acquiring fault-free sub-images of all parts of the railway wagon;
step two: describing expected fault forms by text for each fault-free sub-image;
step three: performing image processing on the failure-free sub-images according to the text description to obtain expected failure images of expected failure forms;
step four: preprocessing the text description to obtain word vectors, encoding all the word vectors to form feature vectors, and obtaining a preprocessed fault image based on the feature vectors; generating a generated fault image of an expected fault form based on the fault-free sub-image and the pre-processed fault image by adopting a generator of the GAN network; then adopting a discriminator of the GAN network to discriminate the expected fault image and the generated fault image until the GAN network reaches a steady state, and obtaining a trained GAN network generator;
step five: performing fault form text description on the image to be processed, and preprocessing the fault form text description to obtain a corresponding preprocessed fault image; and then a trained GAN network generator is adopted to generate a fault image based on the input image to be processed and the corresponding pre-processing fault image, and the fault image is used as training data of the fault detection network.
In the embodiment, given a normal image of a railway wagon part, performing image Processing (PS) according to a text description fault form on the basis of the normal image to obtain an artificial expected fault image; inputting the pre-processed fault image corresponding to the fault-free sub-image and the text description as input, and taking the artificial expected fault image as a generator for outputting a training GAN network; finally, taking the image to be processed and the text description as input to generate a final fault image; the generated fault images can be manually screened, and the available images are reserved and added into the training data set.
The expected failure image obtained in the third step is an image obtained after the artificial PS failure of the failure-free sub-image.
Further, the pretreatment process in the fourth step is the same as that in the fifth step;
the word vector obtaining method in the preprocessing comprises the following steps:
word segmentation processing is carried out on the text description, and the word segmentation processing is used for extracting subsequent word vectors; and obtaining the word vector of each word by a word2vec method.
The method for obtaining the preprocessing fault image in the preprocessing comprises the following steps:
extracting features of all word vectors by adopting an LSTM network, and encoding all words in one description to form feature vectors;
taking the feature vector as primary data with the size of 1x1 and the channel number as feature vector dimension, for example, the feature vector dimension is 64 dimension, and then the primary data is 1x1x 64; carrying out 64 times of convolution operation on the primary data to obtain secondary data with the size of 8x8 and the channel number of 64; and performing deconvolution operation on the two-stage data for a plurality of times to obtain a pre-processed fault image, wherein the size of the pre-processed fault image is the same as that of the non-fault sub-image.
Still further, U-net is employed as a generator in the GAN network. The GAN discriminator is used to determine whether the failure image is PS obtained or generated by the generator.
The input image of the U-net generator also includes a random noise image.
The U-net generator splices (concat) the 3-channel fault-free sub-image, the 1-channel pre-processing fault image and the 1-channel random noise image to obtain a 5-channel primary image, and then the 5-channel primary image is used for generating a generated fault image with an expected fault form. The output of the U-net network is a failure image that is generated using the failure-free image and the literal information.
The input of the GAN arbiter is a failure-free image + text description coding feature vector + an expected failure image obtained by PS or a failure-free image + text description coding feature vector + a generator generated image. The goal of the arbiter is to distinguish between the two cases. The goal of the generator is to make the generated image closer to the PS image, thus spoofing the arbiter.
After training, the generator may generate a corresponding failure image from the input image and the text description. Because random noise is also added to the input, the same input can produce a variety of different results.
And step five, the fault image generated by the image to be processed is filtered and qualified according to a preset standard and then is used as training data of the fault detection network.
Generating and screening fault images:
when generating the fault image, only a generator in the GAN network is needed. The component image and the textual description of the corresponding fault modality are input to a generator, from which the fault image can be generated. After the image is obtained, the generated error or poor quality image can be removed through manual screening. The remaining images may be added to the training set as fault data.
Still further, the obtaining method of the fault-free sub-image of each component in the first step comprises the following steps: and acquiring the railway wagon image by adopting a linear array camera, obtaining the position information of each part of the railway wagon by a target detection method, and obtaining fault-free sub-images of each part according to the position information.
As an example, the failure-free sub-image of each part may take 1000.
In the method, the trained model can quickly generate a large number of fault images in specific forms. The effect similar to that of the manual PS can be achieved for some simple fault forms such as rod breakage, component tilting and the like. The method of the invention has the advantages that the fault image generation speed is far higher than that of the artificial PS, and the method can help to quickly construct a fault data training set, thereby improving the development efficiency of the deep learning model.
As an example, the text description in step two may include brake cylinder dropping and tilting 20 degrees.
The text description in this embodiment means describing various fault forms requiring PS in one sentence. Each non-failure sub-image corresponds to a failure text description.
Although the invention herein has been described with reference to particular embodiments, it is to be understood that these embodiments are merely illustrative of the principles and applications of the present invention. It is therefore to be understood that numerous modifications may be made to the illustrative embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the different dependent claims and the features described herein may be combined in ways other than as described in the original claims. It is also to be understood that features described in connection with separate embodiments may be used in other described embodiments.
Claims (8)
1. A method for generating a railway wagon fault picture based on text description is characterized by comprising the following steps of,
step one: acquiring fault-free sub-images of all parts of the railway wagon;
step two: describing expected fault forms by text for each fault-free sub-image;
step three: performing image processing on the failure-free sub-images according to the text description to obtain expected failure images of expected failure forms;
step four: preprocessing the text description to obtain word vectors, encoding all the word vectors to form feature vectors, and obtaining a preprocessed fault image based on the feature vectors; generating a generated fault image of an expected fault form based on the fault-free sub-image and the pre-processed fault image by adopting a generator of the GAN network; then adopting a discriminator of the GAN network to discriminate the expected fault image and the generated fault image until the GAN network reaches a steady state, and obtaining a trained GAN network generator;
step five: performing fault form text description on the image to be processed, and preprocessing the fault form text description to obtain a corresponding preprocessed fault image; then a trained GAN network generator is adopted to generate a fault image based on the input image to be processed and the corresponding pre-processing fault image, and the fault image is used as training data of a fault detection network;
the GAN network adopts U-net as a generator;
the input image of the U-net generator in step five also includes a random noise image.
2. The method for generating a rail wagon fault picture based on the text description as claimed in claim 1, wherein,
the pretreatment process in the fourth step is the same as that in the fifth step;
the word vector obtaining method in the preprocessing comprises the following steps:
word segmentation processing is carried out on the text description, and word2vec method is adopted to obtain word vectors of each word.
3. The method for generating a rail wagon fault picture based on the text description as claimed in claim 2, wherein,
the method for obtaining the preprocessing fault image in the preprocessing comprises the following steps:
extracting features of all word vectors by adopting an LSTM network, and encoding to form feature vectors;
taking the feature vector as primary data with the size of 1x1 and the channel number as feature vector dimensions, and carrying out convolution operation for 64 times to obtain secondary data with the size of 8x8 and the channel number of 64; and performing deconvolution operation on the two-stage data for a plurality of times to obtain a pre-processed fault image, wherein the size of the pre-processed fault image is the same as that of the non-fault sub-image.
4. A method for generating a rail wagon fault picture based on a text description as claimed in claim 3, wherein,
the U-net generator splices the 3-channel fault-free sub-image, the 1-channel pre-processing fault image and the 1-channel random noise image to obtain a 5-channel primary image, and then the 5-channel primary image is used for generating a generated fault image of an expected fault form.
5. The method for generating a failure picture of a railway wagon based on the text description of claim 4,
and fifthly, screening the fault image generated by the image to be processed according to a preset standard to be qualified and then using the fault image as training data of a fault detection network.
6. The method for generating a rail wagon fault picture based on the text description as claimed in claim 1, wherein,
the obtaining method of the fault-free sub-image of each component in the first step comprises the following steps: and acquiring the railway wagon image by adopting a linear array camera, obtaining the position information of each part of the railway wagon by a target detection method, and obtaining fault-free sub-images of each part according to the position information.
7. The method for generating a failure picture of a railway wagon based on the text description of claim 6, wherein,
the failure-free sub-image of each part is 1000.
8. The method for generating a rail wagon fault picture based on the text description as claimed in claim 1, wherein,
the text description in the second step includes that the brake cylinder is fallen off and inclined by 20 degrees.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310197794.4A CN116452906B (en) | 2023-03-03 | 2023-03-03 | Railway wagon fault picture generation method based on text description |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310197794.4A CN116452906B (en) | 2023-03-03 | 2023-03-03 | Railway wagon fault picture generation method based on text description |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116452906A CN116452906A (en) | 2023-07-18 |
| CN116452906B true CN116452906B (en) | 2024-01-30 |
Family
ID=87129167
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310197794.4A Active CN116452906B (en) | 2023-03-03 | 2023-03-03 | Railway wagon fault picture generation method based on text description |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116452906B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110163267A (en) * | 2019-05-09 | 2019-08-23 | 厦门美图之家科技有限公司 | A kind of method that image generates the training method of model and generates image |
| CN110858269A (en) * | 2018-08-09 | 2020-03-03 | 清华大学 | Criminal name prediction method and device |
| CN111080600A (en) * | 2019-12-12 | 2020-04-28 | 哈尔滨市科佳通用机电股份有限公司 | Fault identification method for split pin on spring supporting plate of railway wagon |
| CN111091555A (en) * | 2019-12-12 | 2020-05-01 | 哈尔滨市科佳通用机电股份有限公司 | Brake shoe breaking target detection method |
| CN111968193A (en) * | 2020-07-28 | 2020-11-20 | 西安工程大学 | Text image generation method based on StackGAN network |
| CN112085734A (en) * | 2020-09-25 | 2020-12-15 | 西安交通大学 | GAN-based image restoration defect detection method |
| CN112699952A (en) * | 2021-01-06 | 2021-04-23 | 哈尔滨市科佳通用机电股份有限公司 | Train fault image amplification method and system based on deep learning |
| CN113362416A (en) * | 2021-07-01 | 2021-09-07 | 中国科学技术大学 | Method for generating image based on text of target detection |
| WO2022131497A1 (en) * | 2020-12-18 | 2022-06-23 | 영남대학교 산학협력단 | Learning apparatus and method for image generation, and image generation apparatus and method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109753998A (en) * | 2018-12-20 | 2019-05-14 | 山东科技大学 | The fault detection method and system, computer program of network are generated based on confrontation type |
-
2023
- 2023-03-03 CN CN202310197794.4A patent/CN116452906B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110858269A (en) * | 2018-08-09 | 2020-03-03 | 清华大学 | Criminal name prediction method and device |
| CN110163267A (en) * | 2019-05-09 | 2019-08-23 | 厦门美图之家科技有限公司 | A kind of method that image generates the training method of model and generates image |
| CN111080600A (en) * | 2019-12-12 | 2020-04-28 | 哈尔滨市科佳通用机电股份有限公司 | Fault identification method for split pin on spring supporting plate of railway wagon |
| CN111091555A (en) * | 2019-12-12 | 2020-05-01 | 哈尔滨市科佳通用机电股份有限公司 | Brake shoe breaking target detection method |
| CN111968193A (en) * | 2020-07-28 | 2020-11-20 | 西安工程大学 | Text image generation method based on StackGAN network |
| CN112085734A (en) * | 2020-09-25 | 2020-12-15 | 西安交通大学 | GAN-based image restoration defect detection method |
| WO2022131497A1 (en) * | 2020-12-18 | 2022-06-23 | 영남대학교 산학협력단 | Learning apparatus and method for image generation, and image generation apparatus and method |
| CN112699952A (en) * | 2021-01-06 | 2021-04-23 | 哈尔滨市科佳通用机电股份有限公司 | Train fault image amplification method and system based on deep learning |
| CN113362416A (en) * | 2021-07-01 | 2021-09-07 | 中国科学技术大学 | Method for generating image based on text of target detection |
Non-Patent Citations (2)
| Title |
|---|
| Optimized GAN for Text-to-Image Synthesis: Hybrid Whale Optimization Algorithm and Dragonfly Algorithm;amsidhar Talasila et al.;《Advances in Engineering Software》;1-11 * |
| 基于深度学习的文本到图像生成方法综述;王宇昊等;《计算机工程与应用》;1-10 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116452906A (en) | 2023-07-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111626300B (en) | Image segmentation method and modeling method of image semantic segmentation model based on context perception | |
| CN111292264B (en) | Image high dynamic range reconstruction method based on deep learning | |
| CN109816032B (en) | Unbiased mapping zero sample classification method and device based on generative countermeasure network | |
| CN112150450B (en) | Image tampering detection method and device based on dual-channel U-Net model | |
| CN113239869B (en) | Two-stage behavior recognition method and system based on key frame sequence and behavior information | |
| CN110766050B (en) | Model generation method, text recognition method, device, equipment and storage medium | |
| CN113343705A (en) | Text semantic based detail preservation image generation method and system | |
| CN109816048B (en) | Image synthesis method based on attribute migration | |
| CN113378815B (en) | Scene text positioning and identifying system and training and identifying method thereof | |
| CN109766918A (en) | Conspicuousness object detecting method based on the fusion of multi-level contextual information | |
| CN114612702A (en) | Image data annotation system and method based on deep learning | |
| CN116452906B (en) | Railway wagon fault picture generation method based on text description | |
| CN115496134B (en) | Traffic scene video description generation method and device based on multi-mode feature fusion | |
| CN112949284A (en) | Text semantic similarity prediction method based on Transformer model | |
| CN116912872A (en) | Drawing identification method, device, equipment and readable storage medium | |
| CN113706572B (en) | End-to-end panoramic image segmentation method based on query vector | |
| CN113344110B (en) | Fuzzy image classification method based on super-resolution reconstruction | |
| CN112990336B (en) | Deep three-dimensional point cloud classification network construction method based on competitive attention fusion | |
| CN115631502A (en) | Character recognition method, character recognition device, model training method, electronic device and medium | |
| CN115438645A (en) | Text data enhancement method and system for sequence labeling task | |
| CN115456941A (en) | Novel electric power insulator defect detection and identification method and system | |
| CN114357984A (en) | Homophone variant processing method based on pinyin | |
| CN116090463A (en) | Business card information extraction system training method and device and storage medium | |
| Kawano et al. | TAG: Guidance-free Open-Vocabulary Semantic Segmentation | |
| CN112951218B (en) | Voice processing method and device based on neural network model and electronic equipment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |