AI Paper Atlas

Research map / Detection and segmentation / CNN features

Learning to Predict Context-adaptive Convolution for Semantic Segmentation

Jianbo Liu, Junjun He, Yu Qiao, Jimmy S. Ren, Hongsheng Li

ECCV 2020 · Detection and segmentation › CNN features · approach: Context modeling for semantic segmentation · 32 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 context-adaptive convolution (CaC) modules to FCN (ablation)FCN ResNet50 · PASCAL Context valmIoU: 45.8 -> 52.0Improved
add global pooling branch (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 52.0 -> 52.5Improved
add deep supervision loss (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 50.7 -> 51.5Improved
add horizontal flip at evaluation (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 51.5 -> 51.8Improved
add multi-scale evaluation (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 51.8 -> 52.5Improved
use input-dependent CaC kernels instead of fixed input-invariant convolutions (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 51.3 -> 52.5Improved
use CaC kernels instead of depth-wise FC predicted kernels (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 51.7 -> 52.5Improved
use CaC kernels instead of kernels predicted from GAP features (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 51.9 -> 52.5Improved
use spatially-varying CaC instead of SE-Head global re-weighting (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 50.9 -> 52.5Improved
use spatially-varying CaC instead of EncHead global re-weighting (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 51.2 -> 52.5Improved
use two CaC modules instead of one (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 51.7 -> 52.5Improved
increase CaC modules from two to four (ablation)CaC-Net ResNet50 · PASCAL Context valmIoU: 52.5 -> 52.1Hurt

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