Research map / Vision-language and multimodal / Human and embodied modeling
Diversify Your Vision Datasets with Automatic Diffusion-based Augmentation
Lisa Dunlap, Alyssa Umino, Han Zhang, Jiezhi Yang, Joseph E. Gonzalez, Trevor Darrell
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 |
|---|---|---|---|
| use ALIA-generated domain prompts instead of user-written prompts (ablation) | ALIA · Waterbirds | class-balanced accuracy (%): 63.64 -> 70.22 | Improved |
| use ALIA-generated domain prompts instead of user-written prompts (ablation) | ALIA · iWildCam subset | accuracy (%): 79.92 -> 82.57 | Improved |
| use ALIA-generated domain prompts instead of user-written prompts (ablation) | ALIA · CUB | accuracy (%): 71.02 -> 71.25 | No clear effect |
| add semantic and confidence-based filtering of edits (ablation) | ALIA · iWildCam subset | accuracy (%): 82.57 -> 84.87 | Improved |
| add semantic and confidence-based filtering of edits (ablation) | ALIA · CUB | accuracy (%): 71.25 -> 72.70 | Improved |
| add semantic and confidence-based filtering of edits (ablation) | ALIA · Waterbirds | class-balanced accuracy (%): 70.22 -> 71.40 | Improved |
| use Img2Img instead of InstructPix2Pix for editing (ablation) | ALIA · iWildCam subset | accuracy (%): 76.23 -> 84.87 | Improved |
| increase number of added generated images beyond 1000 on CUB (ablation) | ALIA · CUB | accuracy | Hurt |
| add text-to-image generated data on CUB | baseline · CUB | accuracy | No clear effect |
Closest papers by meaning
- Effective Data Augmentation With Diffusion Models (ICLR 2024 · similarity 0.89)
- Vision‑Language‑Vision Auto‑Encoder: Scalable Knowledge Distillation from Diffusion Models (NeurIPS 2025 · similarity 0.86)
- Rethinking Few-Shot Adaptation of Vision-Language Models in Two Stages (CVPR 2025 · similarity 0.86)
- Follow-Up Differential Descriptions: Language Models Resolve Ambiguities for Image Classification (ICLR 2024 · similarity 0.86)
- Contrastive Visual Data Augmentation (ICML 2025 · similarity 0.86)
- Multilingual Diversity Improves Vision-Language Representations (NeurIPS 2024 · similarity 0.86)
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.
