What’s the best way to fine-tune an image annotation model? (SAM + Labels project)

I work in the agriculture space and I am currently using SAM on satellite images of farms to segment the tillable versus non-tillable areas within the image.

I want to build a dataset of all of the farmland in the United States that properly labels tillable vs timber vs pasture ground, etc.

SAM brings me closer, but I still need to label the segmented areas myself

I have a dataset that’s been collected over the past 3 years of users …

agriculture annotation computervision dataset image images labels project sam satellite satellite images space united united states work