I have a real world application to detect the slow growth of a cloud image. I'd like to use the Grove Vision AI Module from Seeedstudio. https://www.seeedstudio.com/Grove-Vision-AI-Module-p-5457.html This is a low cost vision AI board with onboad camera and processor. I'm not familar with AI at all and find it very difficult to learn much about it. The terms are all unknown to me and the descriptions about how things work are vague.
It seems most of these devices and software are oriented to determine WHAT an object is. I am only using this for one object (clouds) so that is not nessecary. Also, most use "training" to identify images but I only want to determine incremental growth. Ideally the output would be an image of the cloud with a highlight or marker showing the region of maximum growth over the last 10 to 30 seconds. This is likely to be a change so small as to be not very noticable by a casual observer. Another requirement would be to align each image (maybe 1 per second?) with the previous using the base as the anchor so that growth along the top edge is detected.
Anyone know of any links that might have a similar goal of growth detection?

 

What you're asking for is computer vision with features extraction and edge detection.
Seems interesting but I've not done much with diffuse images like clouds.
https://mommermi.github.io/astronom...ed-cloud-detection-with-machine-learning.html
https://www.google.com/url?sa=t&sou...UQFnoECCYQAQ&usg=AOvVaw2E2vUQE0m7I2r9mMF5K1SZ
Seems possible but not easy.

 

Thanks for the links. Cumulus clouds are not diffuse but very distinct with sharp edges on the top edge of growing clouds. See picture.