1

I have a raster image with building polygons and I want to use these to extract data for deep learning using the ArcGIS Pro tool.

I would like to have a single polygon per image and therefore I'm using the "Labeled Tiles" format.

Additionally, I would like to have a fixed buffer area around each polygon. I don't think the "Fixed Crops" option does this because it crops a fixed size image regardless of the size of the polygon. What I want is to have all the polygon in the image with an added buffer zone around the polygon similarly to what's shown in the image below.

Is there a way to do this?

1 Answer 1

1

This is just an idea, so untested. The Export Training Data For Deep Learning has the parameter input_mask_polygons may be you can create your Bounding mask?

You could create the your bounding boxes by first using the Minimum Bounding Geometry tool then buffer out by 15m. It's that buffered datasets that you supply into the input_mask_polygons parameter.

3
  • Hello, thanks for proposing, it seems like the way to go. I've prepared the buffered dataset as you suggested however I cannot get the Export Training Data to work as I intend. I want each buffered bounding box to be exported into one image and a metadata file containing the object class and the box coordinates. However, it seems to ignore input_mask_polygons and generate chips based on the Tile Size X and Tile Size Y. Can you suggest proper settings for the Export Training Data tool?
    – ali_is
    Oct 7, 2021 at 15:53
  • No, I've not used that parameter before and I too am learning on the job with the new deep learning tools. So if it does not work you need to contact ESRI directly and see if they can offer up a reason why?
    – Hornbydd
    Oct 8, 2021 at 0:13
  • I've managed to make it work. You just provide your buffered polygons as your Input Feature Class and don't use Input Mask Polygons.
    – ali_is
    Oct 11, 2021 at 5:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.