AI Paper Atlas

Research map / Detection and segmentation / CNN features

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

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

ECCV 2018 · Detection and segmentation › CNN features · approach: Context modeling for semantic segmentation · 1,751 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add medium-resolution branch to low-resolution branch (sub4 to sub24) (ablation)ICNet · Cityscapes valmIoU: 59.6 -> 66.5Improved
add high-resolution branch (sub24 to sub124) (ablation)ICNet · Cityscapes valmIoU: 66.5 -> 67.7Improved
use full cascade ICNet instead of half-compressed PSPNet50PSPNet50 compressed to half · Cityscapes valinference time (ms): 170 -> 33Improved
add cascade label guidance (ablation)ICNet with CFF · Cityscapes valmIoU: 66.8 -> 67.7Improved
use CFF instead of 3x3 deconvolution (ablation)ICNet · Cityscapes valmIoU: 66.7 -> 67.7Improved
use CFF instead of 5x5 deconvolution (ablation)ICNet · Cityscapes valmIoU: 66.7 -> 67.7Improved
use CFF instead of 7x7 deconvolution (ablation)ICNet · Cityscapes valinference time (ms): 38 -> 33Improved
use CFF instead of 7x7 deconvolution (ablation)ICNet · Cityscapes valmIoU: 68.0 -> 67.7No clear effect
increase feature downsampling factor from 8 to 32PSPNet50 · Cityscapes valmIoU: 71.7 -> 67.1 mIoU, 446 -> 131 msHurt
keep only 0.25 of kernels (model compression)PSPNet50 · Cityscapes valmIoU: 71.7 -> 59.4 mIoU, 446 -> 72 msHurt
train with both fine and coarse dataICNet · Cityscapes testmIoU: 69.5 -> 70.6Improved
use ICNet instead of PSPNet50 on CamVidPSPNet50 · CamVid testmIoU: 69.1 -> 67.1 mIoU, 185 -> 36 msMixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy