Progressively multi-scale feature fusion for semantic segmentation

doi:10.1016/j.jvcir.2026.104739

Progressively multi-scale feature fusion for semantic segmentation

2026-01-28

0

PRE

AI

DOI：10.1016/j.jvcir.2026.104739

Original

Original request for help

Save

Abstract

En 中文

• We propose a progressive multi-scale feature fusion decoder that achieves precise semantic segmentation with outstanding performance. • It effectively addresses pixel-level misalignment and confusion caused by bilinear upsampling. • The decoder is highly adaptable and versatile, compatible with any backbone network. • It achieves optimal segmentation performance with minimal parameters, resulting in reduced inference time and memory footprint.

Progressively multi-scale feature fusion for semantic segmentation

Abstract

Journal

Organization