AI Paper Atlas

Research map / Detection and segmentation / CNN features

CAR: Class-Aware Regularizations for Semantic Segmentation

Ye Huang, Di Kang, Liang Chen, Xuefei Zhe, Wenjing Jia, Linchao Bao, Xiangjian He

ECCV 2022 · Detection and segmentation › CNN features · approach: Context modeling for semantic segmentation · 20 citations (OpenAlex)

Read the paper

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 intra-class center-to-pixel loss (ablation)ResNet-50 + Self-Attention · Pascal ContextmIoU: 48.32 -> 49.62Improved
add inter-class center-to-center loss (ablation)ResNet-50 + Self-Attention + intra-c2p · Pascal ContextmIoU: 49.17 -> 49.79Improved
add inter-class center-to-pixel loss (ablation)ResNet-50 + Self-Attention + intra-c2p + inter-c2c · Pascal ContextmIoU: 49.79 -> 50.01Improved
add all CAR lossesResNet-50 + Self-Attention · Pascal ContextmIoU: 48.32 -> 50.50Improved
add intra-class center-to-pixel loss (ablation)Swin-Tiny + UperNet · Pascal ContextmIoU: 49.62 -> 49.01Hurt
add inter-class center-to-center loss (ablation)Swin-Tiny + UperNet + intra-c2p · Pascal ContextmIoU: 49.01 -> 50.63Improved
add inter-class center-to-pixel loss (ablation)Swin-Tiny + UperNet + intra-c2p + inter-c2c · Pascal ContextmIoU: 50.63 -> 50.26Hurt
add CARSelf-Attention ResNet-50 · COCO-Stuff10KmIoU: 34.35 -> 36.58Improved
add CARDeepLabV3 ResNet-50 · COCO-Stuff10KmIoU: 34.96 -> 35.13No clear effect
use batch class center instead of image class center (ablation)CAR on ResNet-50 + Self-Attention · Pascal ContextmIoU: 49.78 -> 50.50Improved
use batch class center instead of image class center (ablation)CAR on Swin-Tiny + UperNet · Pascal ContextmIoU: 49.45 -> 50.78Improved
add moving average of class center, decay 0.8 (ablation)CAR on ResNet-50 + Self-Attention · Pascal ContextmIoU: 50.50 -> 49.80Hurt

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