AI Paper Atlas

Research map / Image generation / Image restoration

Disentangled Image Matting

Shaofan Cai, Xiaoshuai Zhang, Haoqiang Fan, Haibin Huang, Jiangyu Liu, Jiaming Liu, Jiaying Liu, Jue Wang, Jian Sun

ICCV 2019 · Image generation › Image restoration · approach: Image harmonization and color transformation · 124 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 the propagation unit (ablation)AdaMatting · Matting module ablationSAD: 44.13 -> 41.70Improved
add the propagation unit (ablation)AdaMatting · Matting module ablationGrad: 17.86 -> 16.89Improved
add the propagation unit (ablation)AdaMatting · Matting module ablationMSE: 0.0111 -> 0.0102Improved
add the trimap-adaptation decoder (ablation)AdaMatting · Matting module ablationSAD: 46.68 -> 41.70Improved
add the trimap-adaptation decoder (ablation)AdaMatting · Matting module ablationMSE: 0.0129 -> 0.0102Improved
add global convolutions (given no sub-pixel and no propagation unit) (ablation)AdaMatting · Matting module ablationSAD: 51.45 -> 45.71Improved
add sub-pixel convolutions (given no global convolutions and no propagation unit) (ablation)AdaMatting · Matting module ablationSAD: 51.45 -> 47.32Improved
add sub-pixel and global convolutions (given no propagation unit) (ablation)AdaMatting · Matting module ablationSAD: 51.45 -> 44.13Improved
use dynamically weighted multi-task loss instead of fixed weights (ablation)AdaMatting · Multi-task loss analysisSAD: 43.54 -> 41.70Improved
use dynamically weighted multi-task loss instead of fixed weights (ablation)AdaMatting · Multi-task loss analysisGrad: 18.35 -> 16.89Improved
use dynamically weighted multi-task loss instead of fixed weights (ablation)AdaMatting · Multi-task loss analysisMSE: 0.0118 -> 0.0102Improved

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