AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Diffusion for Natural Image Matting

Yihan Hu, Yiheng Lin, Wei Wang, Yao Zhao, Yunchao Wei, Humphrey Shi

ECCV 2024 · Image generation › Fast diffusion sampling · approach: Training-free diffusion acceleration · 26 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 UTI self-aligned training to the diffusion matting model (ablation)DiffMatte ViT-S · Composition-1kSAD15: 23.68 -> 20.50Improved
use UTI self-aligned strategy instead of the DDP self-aligned strategy (ablation)DiffMatte ViT-S · Composition-1kSAD1: 21.11 -> 20.61Improved
add diffusion noise rather than iterating one-step ViTMatte with previous prediction as guidance (ablation)ViTMatte · Composition-1kSAD10: 24.84 -> 20.52Improved
decouple lightweight diffusion decoder from encoder (ViTS encoder)DiffMatte · 2048x2048 inferenceinference time per subsequent step: 78% less time, 60% less computeImproved
increase DDIM steps from 1 to 10 (ResNet-34) (ablation)DiffMatte-Res34 · Composition-1kSAD: 31.28 -> 29.20Improved
increase DDIM steps from 1 to 10 (ViT-S) (ablation)DiffMatte-ViTS · Composition-1kSAD: 20.61 -> 20.52No clear effect
increase DDIM steps from 1 to 10 for ViT-B trained on Adobe only (ablation)DiffMatte-ViTB · Distinctions-646SAD: 19.07 -> 19.19Hurt
train on Distinctions-646 instead of Adobe only (ablation)DiffMatte-ViTB · AIM-500SAD (10 steps): 23.57 -> 16.73Improved
use linear instead of cosine or sigmoid noise schedule (ablation)DiffMatte ViT-S · Composition-1kSAD: 21.58 -> 20.52 (cosine to linear)Improved
set input scaling to 0.2 rather than 1.0 (ablation)DiffMatte ViT-S · Composition-1kSAD: 22.19 -> 20.52Improved
increase diffusion decoder channels Nd from 16 to 32 (ablation)DiffMatte ViT-S · Composition-1kSAD: 21.63 -> 20.52Improved
increase Nd from 32 to 48 (ablation)DiffMatte ViT-S · Composition-1kSAD: 20.52 -> 20.29No clear effect

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