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.
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.