Clothing image artistry generation method based on deep learning style migration
A deep learning and artistic technology, applied in neural learning methods, graphic image conversion, image data processing, etc., can solve the problems of single picture style and slow conversion speed, improve viewing interest, good preprocessing operation, and save manpower physical effect
Active Publication Date: 2019-11-22
XIAN UNIV OF TECH
View PDF3 Cites 17 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] The purpose of the present invention is to provide a method for artistically generating clothing images based on deep learning style transfer, which solves the problems of single style and slow conversion speed of existing clothing images
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
[0053] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.
[0054] The present invention is based on the deep learning style transfer method for the artistic generation of clothing images, such as figure 1 shown, including the following steps:
[0055] Step 1: Put the content map and style map into the trained 19-layer VGG network to obtain the content representation and style representation. Among them, the 19-layer VGG network is a known network structure, such as figure 2 As shown, a content diagram of an embodiment, such as image 3 Shown, is the style figure of four different styles in this embodiment (be specifically a, b, c, d four kinds of styles);
[0056] Step 2: Use the filter of the VGG network to encode the content representation and style representation obtained in step 1 to obtain a feature map, and then normalize it through the adaptive specification layer. The specific process of s...
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 invention discloses a clothing image artistry generation method based on deep learning style migration. The clothing image artistry generation method comprises the following steps: step 1, puttinga content graph and a style graph into a trained 19-layer VGG network to obtain content representation and style representation; step 2, encoding the obtained content representation and style representation by using a filter of a VGG network to obtain feature mapping, and normalizing the feature mapping through an adaptive specification layer; step 3, comparing the feature map of the normalized noise picture with the feature map of the content picture and the feature map of the style picture respectively to calculate difference values, and calculating content picture loss, style picture lossand total loss functions respectively; and step 4, training the network according to the obtained total loss function, and generating a result graph through conversion network decoding. According to the clothing image artistry generation method based on deep learning style migration, the problems that an existing clothing image picture is single in style and low in conversion speed are solved.
Description
technical field [0001] The invention belongs to the technical field of image processing and recognition, and in particular relates to a method for artistically generating clothing images based on deep learning style transfer. Background technique [0002] Style transfer refers to the method of transferring the style of an image to another image to generate an image of that style. Traditionally, this problem is mainly solved by texture synthesis and texture migration. Although there are certain effects, the quality of the target image is still not satisfactory. [0003] The traditional image migration algorithm takes random noise as the initial input, and changes the pixel value through iterative optimization to obtain the target result map x', so that the feature expression of this result map is similar to the target feature expression Φ(x), that is, the goal of pixel iteration is Φ (x')≈Φ(x), because each reconstruction result has to undergo multiple iterative optimization...
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): G06T3/00G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06T3/04
Inventor 张九龙马钰玺屈晓娥
Owner XIAN UNIV OF TECH