AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

OntoAug: Rethinking Generative Data Augmentation via Ontology Guidance

Shuo Wang, Zhichuan Wang, Jun Luo

CVPR 2026 · Image generation › Autoregressive and latent generation · approach: Synthetic training data from generators

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 OntoAug generative augmentation versus vanilla trainingResNet-50 (from scratch) · CUBtop-1 accuracy (%): 65.50 -> 84.62Improved
add layout structure augmentation on top of background augmentation (ablation)OntoAug · CUB (ResNet-50)top-1 accuracy (%): 81.53 -> 84.62Improved
add translation (L3) layout alone (ablation)OntoAug background-only · CUB (ResNet-50)top-1 accuracy (%): 81.53 -> 82.29Improved
use generated backgrounds instead of Places backgrounds (ablation)OntoAug · CUB (ResNet-50)top-1 accuracy (%): 77.17 -> 84.62Improved
use diverse background vocabulary instead of habitat-only terms (ablation)OntoAug · Waterbirdsaverage accuracy (%): 73.48 -> 74.82Improved
combine OntoAug with SnapMix instead of Diff-Mix with SnapMixResNet-50 (ImageNet-1K pretrained) · CUBtop-1 accuracy (%): 87.28 -> 88.97Improved
add Diff-Mix to SnapMixResNet-50 (ImageNet-1K pretrained) · CUBtop-1 accuracy (%): 87.75 -> 87.28Hurt
add OntoAug on ViT-B/16ViT-B/16 · CUBaccuracy (%): 89.37 -> 90.78Improved
apply CutMix augmentation in 10-shot CUBResNet-18/ResNet-50/DenseNet-121 · CUB 10-shotaverage test accuracy (%): 31.79 -> 23.36Hurt
add OntoAug augmentation in 10-shot learningthree backbones · CUB 10-shotaverage test accuracy (%): 31.79 -> 67.55Improved
add OntoAug data to GRPO fine-tuningQwen2.5-VL-3B · CUB 8-shotaccuracy: 59.51 -> 66.24Improved
add OntoAug for CAM-based localizationCAM · CUB WSOLIoU50 localization accuracy (%): 50.85 -> 60.54Improved

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