Description
PULSE is a self-supervised photo upsampling algorithm. Instead of starting with the LR image and slowly adding detail, PULSE traverses the high-resolution natural image manifold, searching for images that downscale to the original LR image. This is formalized through the downscaling loss, which guides exploration through the latent space of a generative model. By leveraging properties of high-dimensional Gaussians, the authors aim to restrict the search space to guarantee realistic outputs.
Papers Using This Method
Teach Multimodal LLMs to Comprehend Electrocardiographic Images2024-10-21Detecting Toxic Flow2023-12-10Multi-Head Cross-Attentional PPG and Motion Signal Fusion for Heart Rate Estimation2022-10-14Intermediate Layer Optimization for Inverse Problems using Deep Generative Models2021-02-15PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models2020-03-08