CN114742698A - Wiya line erasing method and device based on depth generation model and storage medium - Google Patents
Wiya line erasing method and device based on depth generation model and storage medium Download PDFInfo
- Publication number
- CN114742698A CN114742698A CN202210433690.4A CN202210433690A CN114742698A CN 114742698 A CN114742698 A CN 114742698A CN 202210433690 A CN202210433690 A CN 202210433690A CN 114742698 A CN114742698 A CN 114742698A
- Authority
- CN
- China
- Prior art keywords
- video
- line
- generation model
- segment
- depth generation
- 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
- 238000000034 method Methods 0.000 title claims abstract description 46
- 230000008569 process Effects 0.000 claims abstract description 12
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 4
- 230000007246 mechanism Effects 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 20
- 238000004590 computer program Methods 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 13
- 239000012634 fragment Substances 0.000 claims description 3
- 230000008439 repair process Effects 0.000 abstract description 4
- 238000012549 training Methods 0.000 abstract description 2
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/04—Context-preserving transformations, e.g. by using an importance map
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/77—Retouching; Inpainting; Scratch removal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20132—Image cropping
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a Wiya line erasing method based on a depth generation model, which comprises the steps of constructing a video depth generation model based on a self-attention mechanism and a convolutional neural network, and training the video depth generation model, or selecting an applicable video depth generation model which is pre-trained; taking any video segment in a target video as a basic erasing unit, utilizing a pre-trained video depth generation model, firstly matching all effective contents in the video segment to obtain effective information to generate reasonable contents, then cutting the generated contents according to the confidence coefficient, and cutting out an area with the confidence coefficient higher than a preset value to cover the area at the position of the Weiya line so as to erase the Weiya line; this process is repeated until all the Weiya lines in all the video segments are erased. The invention uses the relevance between the generated content and other contents and the inter-frame coherence in the video to progressively and iteratively complete video repair segment by segment.
Description
Technical Field
The present invention relates to a video processing method, and more particularly, to a wiegand line erasing method, device and storage medium based on a depth generation model.
Background
At present, in the traditional movie and television play industry, due to the need of plot design, actors usually need to complete actions with high difficulty such as 'cornice and wall walking' by means of Weiya lines. The captured video then requires a special technician to perform wiegand line erasing during post-processing. This process is cumbersome because the technician needs to erase the geodesic lines appearing in the video from outside to inside, frame by frame, by eye.
In recent years, with the development of computing power, deep learning methods represented by deep neural networks are successfully applied to a large number of computer vision tasks and achieve unsophisticated effects including target removal. By means of the deep neural network, a specified area in the video can be automatically and quickly covered by content generated by the deep neural network frame by frame so as to achieve specified target removal. However, most of the existing methods directly use the content generated by the generative model to cover the original content frame by frame at one time, and the reasonable organization and scheduling of the content and the consideration of the inter-frame consistency are lacked. For wiener erasing, due to the viewing requirements of the movie and television drama, wiener erasing not only requires that reasonable content is generated to cover wiener, but also requires that the repaired video has stronger time consistency, and artifacts and color differences are avoided. Therefore, the traditional object removing method is directly applied to the video line wiping field frame by frame to realize automatic line wiping and faces a plurality of problems.
Disclosure of Invention
The invention provides a Wiya line erasing method, a Wiya line erasing device and a Wiya line erasing storage medium based on a depth generation model, which are used for solving the technical problems in the prior art.
The technical scheme adopted by the invention for solving the technical problems in the prior art is as follows: a Wiya line erasing method based on a depth generation model is characterized in that a video depth generation model based on a self-attention mechanism and a convolutional neural network is constructed and trained, or an applicable pre-trained video depth generation model is selected; taking any video segment in a target video as a basic erasing unit, utilizing a pre-trained video depth generation model, firstly matching all effective contents in the video segment to obtain effective information to generate reasonable contents, then cutting the generated contents according to the confidence coefficient, and cutting out an area with the confidence coefficient higher than a preset value to cover the area at the position of the Weiya line so as to erase the Weiya line; this process is repeated until all the Weiya lines in all the video segments are erased.
Further, the method comprises the following specific steps:
firstly, a video clip is intercepted from an original video, and the search of an area to be erased in the video clip and other effective areas in the clip is completed through a pre-trained video depth generation model, so that proper content is generated in a matching manner to cover a Weiya line to be erased in the video clip;
step two, distinguishing the credibility of the generated content by taking the distance between the edge of the generated content and the center of the content as a basis, cutting the generated content according to the credibility, wherein a high-credibility area is used for covering a Weiya line of a corresponding part, and a low-credibility area is discarded;
step three, covering the Wiya line with the cut generated content from outside to inside, and inserting the updated video segment into the original video segment to complete one-time updating;
and step four, repeating the step one to the step three until the Wiya line erasing of the whole video is completed.
Further, the first step comprises the following sub-steps:
a1, sequentially selecting a video clip from a complete video of a Weiya line to be erased;
step A2, selecting an applicable pre-trained video depth generation model for generating content covering a Weiya line;
step A3, for each video frame in the segment, the following processes are performed in sequence: identifying and extracting a Weiya line, performing binarization processing, performing expansion operation and negation operation, and obtaining a frame mask corresponding to each video frame in a fragment;
step A4, normalizing the video clip, multiplying the video clip by the inverse of the mask to obtain the effective area of the video clip, inputting the effective area of the video clip into the video depth generation model, and generating a video clip which is preliminarily erased according to the following formula
Ri’=G[(255-Vi)÷255)×(1-Mi)];
g (-) represents any pre-trained video depth generative model;
t0representing the length of the video segment to be erased;
h represents the height of the video to be erased;
w represents the width of the video to be erased.
Further, the second step comprises the following sub-steps:
step B1, calculating the distance matrix D corresponding to each frame in the video clip according to the following formulai∈{1,…,t0}:
Step B2, according to the given confidence threshold value l, the confidence matrix I corresponding to each frame in the segment is calculated according to the following formulai∈{1,...,t0};
t0Indicating the length of the video sequence to be erased;
a represents a distance matrix DiThe abscissa of any point;
b represents a distance matrix DiThe ordinate of any point;
a' denotes a mask MiThe abscissa of any one of the above non-a coordinates;
b' denotes a mask MiAny of the above is not the ordinate of b.
Further, in step three, the updated video segment is generated according to the following formula:
Vi′=Vi×(1-Mi)+(Mi-Ii)×Ri;
Vi' represents a video segment that completes one iterative update;
Viindicates a length t0An arbitrary video segment of a frame;
Mirepresenting a mask obtained by binarization video-frame-by-video-frame processing;
Iirepresenting a confidence matrix;
Riindicating a video segment that has been erased.
The invention also provides equipment for realizing the Wiya wire erasing method based on the depth generation model, which comprises a memory and a processor, wherein the memory is used for storing a computer program; the processor is configured to execute the computer program and to implement the above-mentioned wiegand line erasing method steps based on the depth generation model when the computer program is executed.
The invention also provides a storage medium, which stores a computer program, and when the computer program is executed by a processor, the Wigner line erasing method based on the depth generation model is realized.
The invention has the advantages and positive effects that: the invention fully utilizes the relevance between the generated content and other contents and the inter-frame continuity in the video and has the following advantages:
the novelty is as follows: the method is used for organizing and taking the content generated by the generation model according to the confidence coefficient for the first time, replaces the traditional one-time repair and frame-by-frame repair, and gradually and iteratively completes video repair segment by segment.
Effectiveness: experiments prove that compared with other existing target removal methods, the intelligent Wiya line erasing method based on the generative model, which is designed by the invention, has improved performance on both the traditional target removal data set and the Wiya line erasing data set, and the effectiveness of the invention is demonstrated.
Universality: the invention mainly focuses on the algorithm angle, is not limited to the generative model, can be used as a plug and play module to be applied to any generative model, and obtains certain performance improvement, thereby indicating that the invention is universal.
Drawings
FIG. 1 is a schematic workflow diagram of a Wiya line erasing method based on a depth generation model according to the present invention.
Fig. 2 is a schematic diagram of the operation steps of obtaining a frame mask corresponding to each video frame in a segment.
Detailed Description
For further understanding of the contents, features and effects of the present invention, the following embodiments are enumerated in conjunction with the accompanying drawings, and the following detailed description is given:
referring to fig. 1 to 2, a wiya line erasure method based on a depth generation model is to construct and train a video depth generation model based on a self-attention mechanism and a convolutional neural network, or select an applicable video depth generation model which is pre-trained; taking any video segment in a target video as a basic erasing unit, utilizing a pre-trained video depth generation model, firstly matching all effective contents in the video segment, acquiring matched effective information to generate reasonable contents, then cutting the generated contents according to the confidence coefficient, and cutting out an area with the confidence coefficient higher than a preset value to cover the position of a Weiya line so as to erase the Weiya line; this process is repeated until all the Weiya lines in all the video segments are erased.
For any video segment, a wiegand line erasing method from outside to inside can be adopted, namely, the wiegand line is gradually covered from the outside of the wiegand line to the inside of the wiegand line, and each video segment can need one or more iterations to be completely erased. The erasing of one video clip is completed, the erasing of the video clip is completed, and then the erasing of the next video clip is started until all the Weiya lines in all the video clips are erased.
The video depth generation model can adopt an applicable video depth generation model in the prior art; or may be implemented by software or components in the prior art and by conventional technical means.
Preferably, a wiya line erasing method based on the depth generation model may include the following specific steps:
step one, a video segment can be cut from an original video, the search of the area to be erased in the video segment and other effective areas in the segment can be completed through a pre-trained video depth generation model, and therefore appropriate content can be generated in a matching mode to cover the Weiya line to be erased in the video segment.
And step two, the distance between the edge of the generated content and the center of the content can be used as a basis for distinguishing the credibility of the generated content, the generated content can be cut according to the credibility, the high-credibility area is used for covering the Weiya line of the corresponding part, and the low-credibility area can be discarded.
And step three, covering the Wiya line with the cut generated content from outside to inside, namely gradually covering the cut generated content from the outside of the Wiya line to the inside of the Wiya line, and then inserting the updated video clip into the original video clip to finish one-time updating.
And step four, repeating the step one to the step three until the Wiya line erasing of the whole video is completed.
Preferably, step one may comprise the sub-steps of:
step A1, sequentially selecting a video clip from a complete video of a Weiya line to be erased;
step A2, selecting an applicable pre-trained video depth generation model for generating content covering a Weiya line;
step a3, for each video frame in the segment, the following processes can be performed in sequence: identifying and extracting a Weiya line, performing binarization processing, performing expansion operation and negation operation, and obtaining a frame mask corresponding to each video frame in a fragment;
step A4, normalizing the video clip, multiplying the normalized video clip by the inverse of the mask to obtain the effective area of the video clip, inputting the effective area of the video clip into the video depth generation model, and generating a video clip with preliminary erasure by the following formula
Ri’=G[(255-Vi)÷255)×(1-Mi)];
g (-) represents any pre-trained video depth generative model;
t0representing the length of the video segment to be erased;
h represents the height of the video to be erased;
w represents the width of the video to be erased.
The Wiya wire identification extraction, binarization processing, expansion operation, negation operation and normalization processing can adopt the applicable modules in the prior art; or may be implemented using software or modules as known in the art and using conventional techniques.
Further, the second step may comprise the following sub-steps:
in step B1, the distance matrix D corresponding to each frame in the video clip can be calculated according to the following formulai∈{1,…,t0}:
Step B2, according to the given confidence threshold value l, the confidence matrix I corresponding to each frame in the segment is calculated according to the following formulai∈{1,...,t0};
The min function represents a function consisting of the minimum values on the common domain among the set of functions of the contained elements.
t0Indicating the length of the video sequence to be erased;
a represents a distance matrix DiThe abscissa of any point;
b represents a distance matrix DiThe ordinate of any point;
a' represents a mask MiThe abscissa of any one of the above other than a;
b' denotes a mask MiThe ordinate of any one of the above non-b;
l represents a confidence threshold.
Preferably, in step three, the updated video segment can be generated according to the following formula:
Vi′=Vi×(1-Mi)+(Mi-Ii)×Ri;
Vi' represents a video segment that completes one iterative update;
Viindicates a length t0An arbitrary video segment of a frame;
Mirepresenting a mask obtained by binarization video frame by video frame processing;
IirepresentA confidence matrix;
Riindicating a video segment that has been erased.
The invention also provides equipment for realizing the Wiya wire erasing method based on the depth generation model, which comprises a memory and a processor, wherein the memory is used for storing a computer program; the processor is configured to execute the computer program and to implement the above-mentioned wiegand line erasing method steps based on the depth generation model when the computer program is executed.
The invention also provides a storage medium, which stores a computer program, and when the computer program is executed by a processor, the Wigner line erasing method based on the depth generation model is realized.
The working process and working principle of the present invention are further explained by a preferred embodiment of the present invention as follows:
the invention relates to a Wiya line erasing method based on a depth generation model, which takes a video clip as a basic erasing unit, utilizes a pre-trained video depth generation model, firstly matches all effective contents in the video clip, and acquires effective information to generate reasonable contents; and organizing the generated content by using an iterative erasure algorithm to gradually complete erasure.
The invention relates to a Wiya line erasing method based on a depth generation model, which mainly comprises two stages: the task of the first stage is to acquire content by using a pre-trained video depth generation model, and the task of the second stage is mainly to organize the generated content to cover the Weiya lines of the original video segment by segment to complete the Weiya line erasing.
The video depth generation model firstly intercepts a video clip from an original video for operation, and the operation process mainly completes the search of an area to be erased in the video clip and other effective areas in the clip, so that proper content is matched for covering a Weiya line to be erased in the video clip. The generated content is cut from the outside and the inside according to a certain thickness, the area with the certain thickness on the outside is regarded as high credibility to cover the Weiya line of the corresponding part, and the area on the inside is regarded as low credibility and discarded. Inserting the updated video segment into the original video to complete one Wiya line erasing, and repeating the steps for a plurality of times until the erasing of the whole video is completed.
The Wiya wire erasing method based on the depth generation model comprises the following specific steps:
1) firstly, a trained deep generative model G (-) is selected for generating contents and a complete video to be repairedThen sampling the complete video, and selecting a video segment
2) As shown in fig. 2, for each video frame in a sampled video clip, a wiya line identification extraction, a binarization process, an expansion operation, and a negation operation are sequentially performed to obtain a frame mask corresponding to each video frame in the clip
3) Video clip ViThe effective area of the video clip obtained by multiplying the normalized video clip by the mask is input into a generation model, and a video clip subjected to coarse erasure is generated through a formula (1)
We defineRepresenting an original video of length t frames,indicates a length t0Any video segment of a frame.Representing the mask resulting from the binarization video-frame-by-video frame processing,representing the results of the line rubbing. G (-) represents any deep generative model that has completed training. Then each iteration process will generate a thread wiping result first:
Ri’=G[(255-Vi)÷255)×(1-Mi)] (1);
4) calculating the distance matrix D corresponding to each frame in the video clip according to the mask of each frame in the clip and the formula (2)i∈{1,…,t0}。
5) According to a given confidence threshold value l, calculating a confidence matrix I corresponding to each frame in the segment according to a formula (3)i∈{1,...,t0}. Then the erasing structure V of the iteration can be obtained by the formula (4)i', will Vi' insert back to V in orderiThe line wiping can be finished.
l represents a confidence threshold;
t0indicating the length of the video sequence to be erased;
a represents a distance matrix DiThe abscissa of any point;
b represents a distance matrix DiThe ordinate of any point;
a' denotes a mask MiThe abscissa of any one of the above non-a coordinates;
b' denotes a mask MiAny of the above is not the ordinate of b.
V is arrangedi' indicate completion of one iterative updateThe video clip of (1); then there are:
Vi′=Vi×(1-Mi)+(Mi-Ii)×Ri (4);
Viindicates a length t0An arbitrary video segment of a frame;
Mirepresenting a mask obtained by binarization video frame by video frame processing;
Iirepresenting a confidence matrix;
Riindicating a video segment that has been erased.
6) Repeating the steps 1) to 5) until all the video segments are erased, and obtaining the erasing result of the video.
The above-mentioned embodiments are only for illustrating the technical ideas and features of the present invention, and the purpose thereof is to enable those skilled in the art to understand the contents of the present invention and to carry out the same, and the present invention shall not be limited to the embodiments, i.e. the equivalent changes or modifications made within the spirit of the present invention shall fall within the scope of the present invention.
Claims (7)
1. A Weiya line erasing method based on a depth generation model is characterized in that a video depth generation model based on a self-attention mechanism and a convolutional neural network is constructed and trained, or an applicable pre-trained video depth generation model is selected; taking any video segment in a target video as a basic erasing unit, utilizing a pre-trained video depth generation model, firstly matching all effective contents in the video segment to obtain effective information to generate reasonable contents, then cutting the generated contents according to the confidence coefficient, and cutting out an area with the confidence coefficient higher than a preset value to cover the area of the Weiya line so as to erase the Weiya line; this process is repeated until all the Weiya lines in all the video segments are erased.
2. The wiya line erasing method based on the depth generation model as claimed in claim 1, comprising the following specific steps:
firstly, a video clip is intercepted from an original video, and the search of an area to be erased in the video clip and other effective areas in the clip is completed through a pre-trained video depth generation model, so that proper content is generated in a matching manner to cover a Weiya line to be erased in the video clip;
step two, distinguishing the credibility of the generated content by taking the distance between the edge of the generated content and the center of the content as a basis, cutting the generated content according to the credibility, wherein a high-credibility area is used for covering a Weiya line of a corresponding part, and a low-credibility area is discarded;
step three, covering the Wiya line with the cut generated content from outside to inside, and inserting the updated video segment into the original video segment to complete one-time updating;
and step four, repeating the step one to the step three until the Wiya line erasing of the whole video is completed.
3. The wiegand line erasing method based on the depth generation model as claimed in claim 2, wherein the first step comprises the following substeps:
step A1, sequentially selecting a video clip from a complete video of a Weiya line to be erased;
step A2, selecting an applicable pre-trained video depth generation model for generating content covering a Weiya line;
step A3, for each video frame in the segment, the following processes are performed in sequence: identifying and extracting a Weiya line, performing binarization processing, performing expansion operation and negation operation, and obtaining a frame mask corresponding to each video frame in a fragment;
step A4, normalizing the video clip, multiplying the video clip by the inverse of the mask to obtain the effective area of the video clip, inputting the effective area of the video clip into the video depth generation model, and generating a video clip which is preliminarily erased according to the following formula
Ri’=G[(255-Vi)÷255)×(1-Mi)];
g (-) represents any pre-trained video depth generative model;
t0representing the length of the video segment to be erased;
h represents the height of the video to be erased;
w represents the width of the video to be erased.
4. The wiya line erase method based on the depth generative model of claim 2, wherein the second step comprises the following substeps:
step B1, calculating the distance matrix D corresponding to each frame in the video clip according to the following formulai∈{1,…,t0}:
Step B2, according to the given confidence threshold l, calculating the confidence matrix I corresponding to each frame in the segment according to the following formulai∈{1,…,t0};
t0Indicating the length of the video sequence to be erased;
a represents a distance matrix DiThe abscissa of any point;
b represents a distance matrix DiThe ordinate of any point;
a' represents a mask MiThe abscissa of any one of the above non-a coordinates;
b' denotes a mask MiAny of the above is not the ordinate of b.
5. The wiya line wipe method based on depth generation model of claim 2 wherein in step three, the updated video segment is generated according to the following formula:
Vi′=Vi×(1-Mi)+(Mi-Ii)×Ri;
Vi' represents a video segment that completes one iterative update;
Viindicates a length t0An arbitrary video segment of a frame;
Mirepresenting a mask obtained by binarization video frame by video frame processing;
Iirepresenting a confidence matrix;
Riindicating a video segment that has been erased.
6. An apparatus for implementing a wiegand line erasure method based on a depth generation model, comprising a memory and a processor, wherein the memory is configured to store a computer program; the processor for executing the computer program and for implementing the wiener line wipe method steps based on the depth generative model as claimed in any one of claims 1 to 5 when the computer program is executed.
7. A storage medium storing a computer program which, when executed by a processor, carries out the wiegand line erasure method steps based on a depth-generating model according to any one of claims 1 to 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210433690.4A CN114742698A (en) | 2022-04-24 | 2022-04-24 | Wiya line erasing method and device based on depth generation model and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210433690.4A CN114742698A (en) | 2022-04-24 | 2022-04-24 | Wiya line erasing method and device based on depth generation model and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114742698A true CN114742698A (en) | 2022-07-12 |
Family
ID=82284384
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210433690.4A Pending CN114742698A (en) | 2022-04-24 | 2022-04-24 | Wiya line erasing method and device based on depth generation model and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114742698A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115439737A (en) * | 2022-10-13 | 2022-12-06 | 哈尔滨市科佳通用机电股份有限公司 | Railway box wagon window fault image identification method based on image restoration |
WO2024183118A1 (en) * | 2023-03-07 | 2024-09-12 | Hong Kong Applied Science and Technology Research Institute Company Limited | Method for optimizing workflow-based neural network including attention layer |
-
2022
- 2022-04-24 CN CN202210433690.4A patent/CN114742698A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115439737A (en) * | 2022-10-13 | 2022-12-06 | 哈尔滨市科佳通用机电股份有限公司 | Railway box wagon window fault image identification method based on image restoration |
WO2024183118A1 (en) * | 2023-03-07 | 2024-09-12 | Hong Kong Applied Science and Technology Research Institute Company Limited | Method for optimizing workflow-based neural network including attention layer |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Liu et al. | Retinex-inspired unrolling with cooperative prior architecture search for low-light image enhancement | |
CN114742698A (en) | Wiya line erasing method and device based on depth generation model and storage medium | |
Zafeiriou et al. | The menpo facial landmark localisation challenge: A step towards the solution | |
US9418280B2 (en) | Image segmentation method and image segmentation device | |
Wu et al. | Deep generative model for image inpainting with local binary pattern learning and spatial attention | |
US11386293B2 (en) | Training image signal processors using intermediate loss functions | |
CN109522950B (en) | Image scoring model training method and device and image scoring method and device | |
CN109918539B (en) | Audio and video mutual retrieval method based on user click behavior | |
CN109815355A (en) | Image search method and device, storage medium, electronic equipment | |
US7522749B2 (en) | Simultaneous optical flow estimation and image segmentation | |
WO2020019591A1 (en) | Method and device used for generating information | |
CN110197183A (en) | A kind of method, apparatus and computer equipment of Image Blind denoising | |
CN110675359A (en) | Defect sample generation method and system for steel coil surface and electronic equipment | |
US20210295016A1 (en) | Living body recognition detection method, medium and electronic device | |
CN110910445A (en) | Object size detection method and device, detection equipment and storage medium | |
CN114266894A (en) | Image segmentation method and device, electronic equipment and storage medium | |
CN112115131A (en) | Data denoising method, device and equipment and computer readable storage medium | |
CN112270747A (en) | Face recognition method and device and electronic equipment | |
CN110795623B (en) | Image enhancement training method and system and computer readable storage medium | |
Zhao et al. | Towards authentic face restoration with iterative diffusion models and beyond | |
Du et al. | Boosting dermatoscopic lesion segmentation via diffusion models with visual and textual prompts | |
CN114626118A (en) | Building indoor model generation method and device | |
CN108681731A (en) | A kind of thyroid cancer ultrasound picture automatic marking method and system | |
CN111325212A (en) | Model training method and device, electronic equipment and computer readable storage medium | |
CN112819755A (en) | Thyroid nodule TI-RADS grading system and method |
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 |