Sentinel-2 Super-Resolution GUI
A simple, user-friendly desktop application to increase the resolution of Sentinel-2 satellite images.
This tool takes the standard 10-meter Blue, Green, Red, and Near-Infrared (NIR) bands and uses a pre-trained AI model to create a higher-resolution, 2x upscaled image. It's designed for users who want to easily enhance their satellite imagery without complex command-line tools.
TL;DR (For the Impatient)
Install stuff: pip install pyqt5 torch safetensors rasterio pillow numpy Get the Model: Download the REC_Real_L1B.safetensors file from Releases. Run the App: python L1BSR-GUI.py Load Data: Select the model file and your four Sentinel-2 band files (B02, B03, B04, B08). The app is meant to be used for Sentinel-2 data downloaded from Copernicus Browser. The band files requirements are:
Image format: TIFF (16-bit)
Image resolution: HIGH (original code asked for band files with less than 500px x 500px resolution but larger resolutions work, albeit slowly)
Coordinate system: UTM (not tested on other systems available in Copernicus Browser but they might work)
Layers (Raw): B02, B03, B04, B08.
Layers (Visualised): True color (optional).
... continue reading