Can I read SG-One: Similarity Guidance for One-Shot Segmentation on EtoBox?

SG-One: Similarity Guidance for One-Shot Segmentation by 19D087 SHANKARMAHADEVAN G is a document available to read on EtoBox.

What is SG-One: Similarity Guidance for One-Shot Segmentation about?

This document proposes a Similarity Guidance Network (SG-One) to tackle the challenging task of one-shot semantic segmentation. SG-One predicts the segmentation mask of a query image with reference to only one densely labeled support image of the same category. It adopts a masked average pooling strategy to obtain robust representative features of the support image. SG-One then leverages cosine similarity to build relationships between support and query image features, guiding the segmentation of objects in

Author

19D087 SHANKARMAHADEVAN G

Language

EN