Shulou(Shulou.com)06/01 Report--

This article shows you how to use AI to transfer clothes between photos, the content is concise and easy to understand, can definitely brighten your eyes, through the detailed introduction of this article, I hope you can learn something.

The AI transfers clothes between photos!

The algorithm represents the pose and shape of the body as a parametric grid, which can be reconstructed from a single image and easily placed.

Given an image of a person, a composite image of that person can be created in a different pose or wearing different clothes obtained from another input image.

Watch the video to see all the examples!

The video is about a new paper about Facebook Reality Labs, which will be presented at the 2020 European Conference on computer Vision (ECCV). At the meeting

Focus on manual re-rendering from a single image.

In short, given an image of a person, a composite image of that person can be created in different poses or different clothes obtained from another input image.

This is called posture transfer and clothing transfer.

Given a specific image of a person, a new image of that person can be synthesized with different target body postures.

This new technology basically consists of four main steps:

Using the DensePose developed in another paper, we can use the correspondence between the input image and the SMPL to extract some of the textures represented in the UV texture map discussed earlier.

Then use FeatureNet, which is the U-Net-like convolution network shown in this figure.

Part of the UV texture map is converted into a complete UV feature map, thus providing a richer representation for each texture pixel.

Finally, use RenderNet, a network of Pix2PixHD-based generators that can be used to convert semantic tag maps into realistic images or to synthesize portraits from facial tag maps.

In this case, Pix2Pix is used on the feature image to generate a realistic image of the person lying down.

The above content is how to use AI to transfer clothes between photos. Have you learned any knowledge or skills? If you want to learn more skills or enrich your knowledge reserve, you are welcome to follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.