AI-based real world super resolution application for React OS, Windows and Linux
Paper "ReactSR: Efficient Real-World Super-Resolution Application in a Single Floppy Disk" was accepted to 20th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISAPP 2025). The article reveals the technical details of training the models underlying ReactSR and comparison with analogues.
- ReactSR supports Windows (XP and later), Linux (tested on current versions of Mint, Ubuntu, RedOS, Alt Linux and Astra Linux) and ReactOS. macOS and Android ARE NOT SUPPORTED.
- ReactSR runs on Windows via pre-installed .NET Framework, on ReactOS via manually installed .NET Framework 4.0, on other systems you should install Mono runtime manually.
- x86, x86-64, Itanium and ARM processors are supported.
| Operating system | Version | Runtime |
|---|---|---|
| Windows | XP, Vista, 7, 8, 8.1, 10, 11, Embedded Standard 7 | .NET Framework 4.0 and higer, Mono 6.12 |
| Linux | Mint, Ubuntu, RedOS, Alt Linux, Astra Linux | Mono 6.12 |
| ReactOS | 0.4.14 | .NET Framework 4.0 |
- OS: Windows 7 Ultimate x64
- Runtime: .NET Framework 4.5
- IDE: SharpDevelop 5.1
- Language: C# 5
- GUI framework: Windows Forms
ReactSR v2.0 is based on [ECCV 2024] SMFANet: A Lightweight Self-Modulation Feature Aggregation Network for Efficient Image Super-Resolution network trained as real-world SR model via RealESRGAN-like pipeline. We use original SMFANet++ pretrained model (https://github.com/Zheng-MJ/SMFANet) to initialize parameters of our model.
