Research map / Image generation / Autoregressive and latent generation
SCalDA: Semantics-Calibrated and Diffusion-Enhanced Data Augmentation
Shibo Lv, Jianmin Jiang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add saliency-based label alignment on top of inpainting (vs random masks with area-based labels) (ablation) | SCalDA · CUB-200-2011 | top-1 accuracy (%): 86.92 -> 87.58 | Improved |
| add metric-based feature alignment (ablation) | SCalDA · CUB-200-2011 | top-1 accuracy (%): 87.58 -> 88.17 | Improved |
| add diffusion inpainting with random masks and area-based labels (ablation) | Vanilla ResNet-50 · CUB-200-2011 | top-1 accuracy (%): 86.64 -> 86.92 | No clear effect |
| replace area-ratio labels and random masks with saliency-guided masks and saliency labels plus inpainting (ablation) | Vanilla ResNet-50 · CUB-200-2011 | top-1 accuracy (%): 86.64 -> 87.58 | Improved |
| use full SCalDA instead of Vanilla | ResNet-50 · CUB-200-2011 | top-1 accuracy (%): 86.64 -> 88.17 | Improved |
| use SCalDA instead of DiffMix in few-shot | ResNet-50 · CUB 2-shot | top-1 accuracy (%): 36.19 -> 40.21 | Improved |
| use SCalDA instead of DiffMix in few-shot | ResNet-50 · CUB 20-shot | top-1 accuracy (%): 80.96 -> 84.16 | Improved |
| use SCalDA instead of Vanilla on general datasets | PreActResNet-18 · CIFAR-100 | top-1 accuracy (%): 76.33 -> 83.18 | Improved |
| use SCalDA instead of DiffuseMix on general datasets | PreActResNet-18 · Tiny-ImageNet-200 | top-1 accuracy (%): 65.77 -> 65.89 | No clear effect |
| use SCalDA instead of SGD-Mix | ResNet-50 · FGVC average | top-1 accuracy (%): 90.28 -> 91.10 | Improved |
| use SCalDA instead of PuzzleMix for label-image alignment | CutMix/PuzzleMix · Saliency Alignment Error | SAE (%): 38.61 -> 29.72 | Improved |
| use SCalDA instead of PuzzleMix for preserved saliency | PuzzleMix · Mixed Saliency | Mixed Saliency: 1.23 -> 1.41 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DiffuseMix — DiffuseMix: Label-Preserving Data Augmentation with Diffusion Models (CVPR 2024) · better on 2 of 2 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
- Semantic-Guided Generative Image Augmentation Method with Diffusion Models for Image Classification (AAAI 2024 · similarity 0.86)
- Effective Data Augmentation With Diffusion Models (ICLR 2024 · similarity 0.85)
- $\mathbf{D^{3}S^{2}}$: Diffusion-Guided Dataset Distillation for Semantic Segmentation (NeurIPS 2026 · similarity 0.85)
- Data Augmentation with Diffusion for Open-Set Semi-Supervised Learning (NeurIPS 2024 · similarity 0.84)
- An Augmentation-Aware Theory for Self-Supervised Contrastive Learning (ICML 2025 · similarity 0.83)
- Beyond Words: Augmenting Discriminative Richness via Diffusions in Unsupervised Prompt Learning (CVPR 2025 · similarity 0.83)
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.
