Hi I’m a GIS professional and I am generally new to this space. I saw esri put out a model that takes text and detects+segments objects in an image using a combination of GroundingDINO and SAM. My use case is aerial imagery of 15cm. Here is their model: TextSAM
However I’m not satisfied with the results and I think it may be because the model (GroundingDINO) wasn’t trained sufficiently with aerial imagery for my use case.
Are there open source aerial imagery, RGB ~15cm resolution that I can use to fine-tune this model? Preferably with text captions for each image as well.
Thank you