Built independently by an author, for readers. Read the story and support ChapterPal
ChapterPal Logo
Log inSign up

keyword

cascade feature fusion

Cascade feature fusion is a neural network architectural mechanism that progressively combines feature representations extracted from different spatial resolutions and abstraction levels in a sequential, step-by-step manner. In this process, deep and semantically rich feature maps generated from lower-resolution processing branches are iteratively upsampled and merged with finer, high-resolution spatial features from parallel or earlier network stages. This staged integration allows computer vision models to capture high-level contextual information alongside precise boundary and spatial details while maintaining computational efficiency. Intermediate fusion modules within this design often utilize auxiliary label supervision to align multi-scale features and promote effective gradient propagation during training.

1 item

ICNet for Real-Time Semantic Segmentation on High-Resolution Images

ICNet for Real-Time Semantic Segmentation on High-Resolution Images

Hengshuang Zhao, Xiaojuan Qi, Xiaoyong Shen, Jianping Shi, Jiaya Jia

OrganizationsSenseTimeTencentThe Chinese University of Hong Kong

Why you should read this

Introduces an image cascade network that combines multi-resolution branches with cascade feature fusion to deliver real-time, high-accuracy semantic segmentation on high-resolution images.

We focus on the challenging task of real-time semantic segmentation in this paper. It finds many practical applications and yet is with fundamental difficulty of reducing a large portion of computation for pixel-wise label inference. We propose an image cascade network (ICNet) that incorporates multi-resolution branches under proper label guidance to address this challenge. We provide in-depth analysis of our framework and introduce the cascade feature fusion unit to quickly achieve high-quality segmentation. Our system yields real-time inference on a single GPU card with decent quality results evaluated on challenging datasets like Cityscapes, CamVid and COCO-Stuff.

Added

2026-09-24