AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

Unsupervised Semantic Correspondence Using Stable Diffusion

Eric Hedlin, Gopal Sharma, Shweta Mahajan, Hossam Isack, Abhishek Kar, Andrea Tagliasacchi, Kwang Moo Yi

NeurIPS 2023 · Image generation › Text-to-image diffusion · approach: Compositional text-to-image guidance · 22 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 random crops during embedding optimization (ablation)Stable Diffusion correspondence method · PF-WillowPCK@0.05: 45.5 -> 53.0Improved
average attention over multiple U-Net layers instead of a single layer (ablation)Stable Diffusion correspondence method · PF-WillowPCK@0.05Improved
optimize more embeddings from random initializations and average (ablation)Stable Diffusion correspondence method · PF-WillowPCK@0.05Improved
average attention over more crops at inference (ablation)Stable Diffusion correspondence method · PF-WillowPCK@0.05Improved
use optimized prompt-embedding attention instead of DINO+NN featuresunsupervised matching · SPair-71kPCK@0.1: 33.3 -> 45.4Improved
unsupervised method vs strongly supervised VATOur method · PF-WillowPCK@0.05 / PCK@0.1: PCK@0.1 81.6 -> 84.3; PCK@0.05 52.8 -> 53.0Mixed

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