Jan 05, 2020 · To achieve this, they first collect an unpaired dataset and introduce a way to synthesize paired training data for self-supervised learning. Then, to unselfie a photo, they propose a new three-stage pipeline, where they first find a target neutral pose, inpaint the body texture, and finally refine and composite the person on the background.
pulseの8倍画像も他と比べて細部まで明確に描かれていますが、pulseの64倍画像はさらに詳細です。 rudin氏によると、pulseは他のツールには行えないような「低品質の画像からリアルな写真を作り出す」ことが可能とのこと。 PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models  Unsupervised Translation of Programming Languages  PIFuHD: Multi-Level Pixel-Aligned Implicit Function for High-Resolution 3D Human Digitization  High-Resolution Neural Face Swapping for Visual Effects 
The latent space contains a compressed representation of the image, which is the only information the decoder is allowed to use to try to reconstruct the input as faithfully as possible. Now that we know what level of detail the model is capable of extracting, we can probe the structure of the latent space.Jun 14, 2020 · Publication: "PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models," Sachit Menon, Alexandru Damian, Shijia Hu, Nikhil Ravi, Cynthia Rudin. IEEE/ CVF International Conference on Computer Vision and Pattern Recognition (CVPR), June 14-19, 2020. arXiv:2003.03808. on June 14, 2020.