Rescaling Color Channels for Keras

  • 29 February 2024
  • 3 replies

Badge +1

This question is not about visualization so it may not be related to
even there are some similarities. My issue is that I am feeding a Keras model where I need to basically use a 0.0 to 1.0 color space. Now for regular RGB images I would simply do a rescaling with /255. For Sentinel2 I’d do /10000 but what is the correct factor for Planet
Asset types

I do not want an individual range per scene but a “standard” rescaling factor. If there are outliers (as with Sentinel2 NIR channel) this would be Ok.
thank you very much for your support.


Best answer by elyhienrich 5 March 2024, 18:23

View original

3 replies

Userlevel 6
Badge +8

Hi @fricke_hs-nb de

Based on our investigation, support will be able to help you with this question. We’ve created a support ticket on your behalf and copied the email that you used in your community account. 

Let us know if you have any further questions. 

Have a nice day! 

Userlevel 7
Badge +13

Some helpful articles that may help others: 

Planet Imagery and Color Correction

Making Sense of Satellite Data, an Open Source Workflow: Color Correction with GIMP

Badge +1

Thank you very much for supporting me. The article about “Planet Imagery and Color Correction” in the section of “Processing for analytic products” indeed gave the needed bit of information.