Inquiry about Super-Resolution Parameter Settings using Deepness Plugin in QGIS

[omranamini]

I am currently working on my thesis and using satellite imagery with a GSD of 74 cm and an approximate resolution of 120 cm. The image is a pansharpened 4-band (RGB + NIR) image with 16-bit depth. I plan to enhance its spatial resolution using the Deepness plugin in QGIS for super-resolution processing.

I have a few questions regarding the parameter settings:

How should I configure the super-resolution parameters to achieve optimal results for my data?

On your website, you mention the importance of data preparation — could you please explain how I should prepare my imagery before running the model?

The model I intend to use (for example, the 2× super-resolution model) states that it was trained on data with 10 cm resolution. Should I rescale my imagery to 10 cm resolution before applying the model?

Since my image is 16-bit, do I need to convert it to 8-bit for compatibility or better performance?

I would greatly appreciate your guidance on how to properly adjust the parameters and prepare my data for the best possible outcome.

[przemyslaw-aszkowski]

Hi,
If you intend to use the model in model ZOO, it will work well only with images of the same type and parameters as the data it was trained on. So the same resolution is needed. There is not much to configure in the superresolution model itself.
Moreover, I think the superresolution model in ZOO is for RGB only, so I'm not sure if it will work with 4-band image. You may need to retrain own model, which is a more complicated task. The image depth (8-16 bit) shouldn't matter I think.