AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

LayoutTransformer: Scene Layout Generation With Conceptual and Spatial Diversity

Cheng-Fu Yang, Wan-Cyuan Fan, Fu-En Yang, Yu-Chiang Frank Wang

CVPR 2021 · Detection and segmentation › Open-set and few-shot detection · approach: Object-level representation learning · 39 citations (OpenAlex)

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
Use LT-Net instead of CanonicalSg2ImScene-graph-to-layout generation · COCO-Stuff testmIOU: 0.42 ±0.01 -> 0.49 ±0.03Improved
Use LT-Net instead of CanonicalSg2ImScene-graph-to-layout generation · COCO-Stuff testRelation accuracy: 50.12 ±0.21 -> 51.36 ±0.45Improved
Use LT-Net instead of Sg2ImLayout-to-image generation · COCO-Stuff testFID: 69.2 ±0.4 -> 76.8 ±0.3Hurt
Use LT-Net instead of CanonicalSg2ImScene-graph-to-layout generation · VG-MSDN testmIOU: 0.174 ±0.025 -> 0.183 ±0.036Improved
Use LT-Net instead of Sg2ImLayout-to-image generation · VG-MSDN testFID: 92.1 ±1.7 -> 99.8 ±1.7Hurt
Add object ID embedding (ablation)LT-Net predictor · COCO-Stuff masked-word predictionMasked accuracy: 74.45 ±0.52 -> 85.68 ±0.21Improved
Add PoP ID embedding (ablation)LT-Net predictor · COCO-Stuff masked-word predictionMasked accuracy: 74.45 ±0.52 -> 75.73 ±0.08Improved
Add object ID and PoP ID embeddings (ablation)LT-Net predictor · COCO-Stuff masked-word predictionMasked accuracy: 74.45 ±0.52 -> 87.12 ±0.17Improved
Add relation consistency loss (ablation)LT-Net layout generator · COCO-Stuff testCoarse-box mIOU: 0.43 ±0.01 -> 0.45 ±0.02Improved
Add relation consistency loss (ablation)LT-Net layout generator · COCO-Stuff testDiversity score: 0.54 ±0.03 -> 0.53 ±0.06No clear effect
Add visual-textual co-attention refinement (ablation)LT-Net layout generator · COCO-Stuff testFID: 85.6 ±0.3 -> 75.3 ±0.8Improved
Add GMM confidence weighting to co-attention (ablation)LT-Net layout generator · COCO-Stuff testFID: 75.3 ±0.8 -> 76.8 ±0.3Hurt

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