Image style migration method and application thereof
An image and style technology, applied in image style transfer and its application field, can solve the problems of cGAN network constraints and constraints, and achieve the effect of reducing the accuracy of the model and saving a lot of consumption.
Pending Publication Date: 2022-08-09
SHANGHAI TECH UNIV
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, due to the special structure of the RAW image, such as the RGB staggered Bayer array, such constraints are difficult to constrain in the loss function of the cGAN network
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0026] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these examples are only used to illustrate the present invention and not to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of the present application.
[0027] The present invention needs to solve the following three technical problems: firstly, the existing color data sets are difficult to meet the needs of reconstructing ISP for computer vision tasks; secondly, in the application of on-chip computing, the mismatch between training data and test data etc. will reduce the accuracy of the model; finally, in order to solve the above two problems, a Bayer dataset is necessary, but ...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The technical scheme of the invention provides an image style migration method, which uses a conditional generative adversarial network to carry out image style migration, and is characterized in that when the conditional generative adversarial network is trained, an image generated by a generator G and an original input image are processed by using a group of reversible micro-transformations and then are put into a discriminator D; therefore, any differentiable and reversible transformation is added into the training process of the conditional generative adversarial network as a constraint which cannot be formulated. Another technical scheme of the invention is to provide an application of the image style migration method. Technicians in the field can apply the data set generated by the method provided by the invention to fine tuning of the model weight for on-chip calculation so as to improve model accuracy reduction caused by misalignment between training data and test data.
Description
technical field [0001] The invention relates to an image style transfer method and its application in image conversion, especially for converting mainstream image data sets to Bayer domain. Background technique [0002] Nowadays, computer vision algorithms based on deep learning have achieved outstanding results in many fields, and in most computer vision applications, cameras are mainly used as information collection devices. Most of the current commercial cameras use a single-sensor solution, that is, a color filter array (CFA) is covered on the sensor surface, and then a color image is obtained through a series of image processing pipelines (ISP). [0003] Therefore, almost all computer vision algorithms take color images as input, but ISP is a processing link designed for human vision, not computer vision tasks, and the existence of ISP may affect subsequent vision tasks. There is a potential impact, and the pixel value of the RAW image is linearly related to the irradi...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06T3/40G06T5/00G06N3/04G06N3/08
CPCG06T3/4015G06N3/08G06T2207/10024G06T2207/20081G06T2207/20084G06N3/045G06T5/73
Inventor 周巍娄鑫
Owner SHANGHAI TECH UNIV