3
$\begingroup$

High-resolution representations are essential for position-sensitive vision problems, such as human pose estimation, semantic segmentation, and object detection.
From HR-Nets we get semantically richer and spatially more precise resultant, what does this statement mean in the paper Deep High Resolution Representation Learning for Visual Recognition.

$\endgroup$
1
  • $\begingroup$ Could you review my answer? $\endgroup$
    – Royi
    Nov 24, 2023 at 14:42

1 Answer 1

1
$\begingroup$

The model task is semantic segmentation.
So for each pixel we assign a class.

In this context:

  1. Semantically Richer
    Being able to detect more classes accurately.
  2. More Precise
    Means the boundary between classes instances is more accurate. So spatially the segmentation is more accurate.

To give a concrete example:

enter image description here

Richer means you can see more colors, as more classes are identified.
Precise means the edges between the classes matches the edges in the real world image.

$\endgroup$

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.