Startup Idea: Automated satellite imagery sorting system

Startup idea to develop a software that automates the process of sorting through satellite images, such as Landsat imagery, to separate useful ones (with little to no cloud cover) from unusable ones (either due to visual glitches and significant cloud cover, typically over 60%). The software should leverage existing cloud masking models to automate this process, enabling it to run on an entire folder of images and eliminate the need for hours of manual sorting.

I work in Earth Observations. As part of my job I have had to manually sort through many thousands of downloaded satellite images such as Landsat imagery to separate out useful ones (little to no cloud cover) from ones that cannot be used (either visual glitches and ones with over a specified percentage, usually 60%, of cloud cover).

Whilst we have methods for detecting and quantifying clouds (cloud masking models), we have not yet automated this process so it can be left to run on an entire folder of images to help narrow down useful ones without it taking hours of manual sorting.