0

I want to detect Land mines in satellite Images. Initially I built a model with each image having multiple labels and trained it to classify the images.

However I want to use Image Segmentation technique as mentioned here : https://towardsdatascience.com/dstl-satellite-imagery-contest-on-kaggle-2f3ef7b8ac40

I downloaded the required images through aws s3 bucket. I want to label each pixel of the multispectral image I have generated from Band files.

However I am facing difficulty in how to labelling.

Are there any open source or otherwise tools to do the same.

EDIT : The images are 12 band multispectral satellite images.

user2777
  • 3
  • 4

1 Answers1

-1

You can use AWS Ground Truth to create a job that can label the images you require. AWS also has released which might help https://aws.amazon.com/about-aws/whats-new/2019/12/amazon-sagemaker-ground-truth-adds-auto-segment-feature-for-semantic-segmentation-labeling/

David Webster
  • 2,208
  • 1
  • 16
  • 27
  • Since the images are satellite images, is there any way of labelling the pixels using only the numpy array of image? – user2777 Oct 28 '20 at 08:36