Super Resolution Generative Adversarial Network
Super-resolution generative adversarial networks (SRGANs) are deep learning models designed to enhance the resolution of images and other data, overcoming limitations of low-resolution input. Current research focuses on improving SRGAN architectures, such as incorporating temporal information for video processing, leveraging enhanced models like ESRGAN and exploring alternative architectures like MLP-Mixers for specific data types (e.g., MRI scans). These advancements have significant implications across diverse fields, including medical imaging (improving diagnostic accuracy), remote sensing (enhancing data from sensors), and engineering (improving the resolution of seismic data and engineering drawings), ultimately leading to more efficient data storage and analysis.
Papers
Applying Physics-Informed Enhanced Super-Resolution Generative Adversarial Networks to Turbulent Non-Premixed Combustion on Non-Uniform Meshes and Demonstration of an Accelerated Simulation Workflow
Mathis Bode
Applying Physics-Informed Enhanced Super-Resolution Generative Adversarial Networks to Finite-Rate-Chemistry Flows and Predicting Lean Premixed Gas Turbine Combustors
Mathis Bode
Applying Physics-Informed Enhanced Super-Resolution Generative Adversarial Networks to Turbulent Premixed Combustion and Engine-like Flame Kernel Direct Numerical Simulation Data
Mathis Bode, Michael Gauding, Dominik Goeb, Tobias Falkenstein, Heinz Pitsch