I am trying to identify and classify marine litter from drone RGB images at beaches. Due to the small size of litter and only having RGB images available, my supervisor recommended to carry out contrast enhancement. Does this actually help to get better image segmentation and classification? I have read that it only changes the display view, so I am not sure how this could somehow help.

  • I am not an expert on this field but here are my two cents. I've seen several examples of image-classification (especially images including water bodies) where they stretch the image or apply a filter to increase contrast between different features. For instance, check this example. I think stretching an image or applying a filter does not change only the display, as it changes the underlying RGB values as well. In conclusion, it can help. Commented Jul 8, 2019 at 13:49


Your Answer

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