AI Paper Atlas

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

RetrieveGAN: Image Synthesis via Differentiable Patch Retrieval

Hung-Yu Tseng, Hsin-Ying Lee, Lu Jiang, Ming-Hsuan Yang, Weilong Yang

ECCV 2020 · Image generation › Text-to-image diffusion · approach: Layout-to-image control · 36 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 ground-truth selection loss (without co-occurrence loss) (ablation)RetrieveGAN · COCO-StuffFID: 56.8 -> 47.8Improved
add ground-truth selection loss (without co-occurrence loss) (ablation)RetrieveGAN · COCO-StuffIS: 8.8 -> 9.7Improved
add co-occurrence loss (without GT selection loss) (ablation)RetrieveGAN · COCO-StuffFID: 56.8 -> 52.8Improved
add co-occurrence loss (without GT selection loss) (ablation)RetrieveGAN · COCO-StuffIS: 8.8 -> 9.8Improved
add both selection losses (ablation)RetrieveGAN · COCO-StuffFID: 56.8 -> 43.2Improved
add both selection losses (ablation)RetrieveGAN · COCO-StuffIS: 8.8 -> 10.6Improved
add both selection losses (ablation)RetrieveGAN · COCO-StuffDS: 0.30 -> 0.34Mixed
use ground-truth bounding boxes instead of predicted onesRetrieveGAN · Visual GenomeFID: 70.3 -> 46.3Improved
use ground-truth bounding boxes instead of predicted onesRetrieveGAN · COCO-StuffFID: 43.2 -> 42.7No clear effect
learned differentiable retrieval instead of pre-defined embedding retrieval (vs PasteGAN)RetrieveGAN · COCO-Stuff (predicted boxes)FID: 59.8 -> 43.2Improved

Later papers that compared with this method

1 later accepted paper reports RetrieveGAN as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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