question

deskcheck1-0579 avatar image
0 Votes"
deskcheck1-0579 asked GiftA-MSFT commented

Image Classification using ML.NET deep learning module

I have not tried image classification using ML.NET. But first off, can I use ML.NET to generate multiple polygons if I "feed" it high-resolution images for training data?

It seems like all it does is output some predictions in text. Can it not output shapefiles, json, or actual polygons, lines, etc?

Maybe a dumb question. But appreciate any clarification.

dotnet-ml-big-data
· 5
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

Hi, yes, the model classifies the image and outputs predictions as text. Can you clarify why you'd need shapefiles as output for prediction?

0 Votes 0 ·

I need to identify field boundaries (meaning polygons) based on images (e.g., if I overlay the model output on Bing Map or Google Map, the polygons align with the map). I need this for research we conduct on soil conservation practices that I need for my GIS processing.

That's not the only application, if ML.NET can swing it. It would also enable us to identify actual structural conservation practices that are present in the field (by feeding the model the latest actual satellite images. (Actually, there are more uses than I can state here).

In order to do this, I need actual image outputs that show the polygons the model identified based on the images that I trained it with. That's the ultimate goal.

Does that make sense?

0 Votes 0 ·

Here's one paper that outlines the usage for deep learning that I mentioned:

https://www.sciencedirect.com/science/article/pii/S2352938519301831

0 Votes 0 ·

Thanks for the information. This question is best suited for mlnet community, I went ahead to add appropriate channel tags.

0 Votes 0 ·

0 Answers