Generative adversarial network-based pixel-level portrait cutout method

Abstract

The invention discloses a generative adversarial network-based pixel-level portrait cutout method and solves the problem that massive data sets with huge making costs are needed to train and optimizea network in the field of machine cutout. The method comprises the steps of presetting a generative network and a judgment network of an adversarial learning mode, wherein the generative network is adeep neural network with a jump connection; inputting a real image containing a portrait to the generative network for outputting a person and scene segmentation image; inputting first and second image pairs to the judgment network for outputting a judgment probability, and determining loss functions of the generative network and the judgment network; according to minimization of the values of theloss functions of the two networks, adjusting configuration parameters of the two networks to finish training of the generative network; and inputting a test image to the trained generative network for generating the person and scene segmentation image, randomizing the generated image, and finally inputting a probability matrix to a conditional random field for further optimization. According tothe method, a training image quantity is reduced in batches; and the efficiency and the segmentation precision are improved.

Description

technical field [0001] The present invention relates to the field of computer vision technology, in particular to a pixel-level portrait matting method, in particular to a pixel-level portrait matting method based on a generative confrontation network, which is used to separate portraits from backgrounds. Background technique [0002] portrait Figure 1 It has always been a hot issue in the field of computer vision. Pixel-level portrait matting requires accurately extracting the foreground of an object from the background, which belongs to a more refined binary semantic segmentation problem. [0003] With the rapid development of e-commerce, portrait matting has a very wide range of application scenarios. For example, more and more people choose to buy clothing on the Internet, so the e-commerce function of searching for things by image came into being. It is very difficult to accurately search for similar clothing, so it is necessary to segment the portraits in the picture...

AI Technical Summary

Problems solved by technology

But because of its simple network structure, it often requires a lot of training set images to train the network

In addition, the data sets used for pixel-level portrait matting are relatively small, and the calibration and production costs of the data sets are huge. A training labeled image sample requires half an hour of manual calibration. However, the segmentation model based on the full convolutional network requires Tens of thousands of images in the training set can achieve good results. It can be seen that manual labeling to obtain available data sets is time-consuming and labor-intensive

Images