Scoring and model training method and device and electronic equipment
A technology of model training and electronic equipment, applied in the field of data processing, can solve the problems of high labor cost, low efficiency, poor aesthetic model effect, etc., and achieve the effect of improving accuracy, improving efficiency and reducing cost
Pending Publication Date: 2022-06-24
BEIJING SOGOU TECHNOLOGY DEVELOPMENT CO LTD
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Among them, there are many ways to train the aesthetic model. One is to train based on the existing aesthetic image dataset (including images and corresponding aesthetic scoring labels); however, the categories of images in the aesthetic image dataset are limited, so that the trained aestheti
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
[0066] In order to make the above objects, features and advantages of the present invention more clearly understood, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.
[0067] One of the core concepts of the embodiments of the present invention is to construct a training set, and use the method of comparing the aesthetic quality of each pair of images in the training set to annotate the images in the image training set; professional aesthetic knowledge is not required, thereby reducing the need for professional aesthetics. The cost of model training improves the efficiency of model training. Then, the annotated training set is used to train the aesthetic model, which can improve the accuracy of the aesthetic score of the aesthetic model compared with the prior art using the existing aesthetic image dataset with limited categories to train the aesthetic model.
[0068] refer to figure 1 , showi...
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 embodiment of the invention provides a scoring and model training method and device and electronic equipment, and the method comprises the steps: collecting a plurality of training images; dividing the plurality of training images into a plurality of image groups, each group of which comprises two training images, and obtaining an aesthetic quality comparison result of the two training images in each image group; according to the aesthetic quality comparison result, marking an image label for each corresponding image group; wherein the image search terms corresponding to the two training images in each image group are the same; and training the aesthetic model by adopting each image group and the image tag of each image group. Therefore, the training image can be labeled without aesthetic scoring of the training image by a professional aesthetic knowledge user, so that the cost of model training can be reduced, and the efficiency of model training can be improved; and the aesthetic model is trained by adopting the created training set, so that the aesthetic scoring accuracy of the aesthetic model can be improved.
Description
technical field [0001] The present invention relates to the technical field of data processing, in particular to a scoring and model training method, device and electronic device. Background technique [0002] Image aesthetic quality can refer to a measure of visually perceived image beauty, which measures the visual appeal of an image to human eyes. Since visual aesthetics is a subjective attribute that tends to involve emotion and personal taste, this makes assessing the aesthetic quality of images a very subjective task; however, there is often a consensus that some images are more visually appealing than others more attractive. Therefore, computational techniques can be used to predict human emotional responses to visual stimuli, so that computers can imitate human aesthetic processes to evaluate the aesthetic quality of images. There are various applications for evaluating the aesthetic quality of images, such as aesthetic-assisted image search, automatic photo enhanc...
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 IPC IPC(8): G06F16/535G06F16/538G06F16/953G06K9/62G06N3/08
CPCG06F16/535G06F16/953G06F16/538G06N3/084G06F18/214
Inventor 李玉杰谢泽华周泽南陈炜鹏许静芳
Owner BEIJING SOGOU TECHNOLOGY DEVELOPMENT CO LTD