REVIEW 3 cited by
Diversify, Don't Fine-Tune: Scaling Up Visual Recognition Training with Synthetic Images
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
read the original abstract
Recent advances in generative deep learning have enabled the creation of high-quality synthetic images in text-to-image generation. Prior work shows that fine-tuning a pretrained diffusion model on ImageNet and generating synthetic training images from the finetuned model can enhance an ImageNet classifier's performance. However, performance degrades as synthetic images outnumber real ones. In this paper, we explore whether generative fine-tuning is essential for this improvement and whether it is possible to further scale up training using more synthetic data. We present a new framework leveraging off-the-shelf generative models to generate synthetic training images, addressing multiple challenges: class name ambiguity, lack of diversity in naive prompts, and domain shifts. Specifically, we leverage large language models (LLMs) and CLIP to resolve class name ambiguity. To diversify images, we propose contextualized diversification (CD) and stylized diversification (SD) methods, also prompted by LLMs. Finally, to mitigate domain shifts, we leverage domain adaptation techniques with auxiliary batch normalization for synthetic images. Our framework consistently enhances recognition model performance with more synthetic data, up to 6x of original ImageNet size showcasing the potential of synthetic data for improved recognition models and strong out-of-domain generalization.
Forward citations
Cited by 3 Pith papers
-
LoFT: LoRA-fused Training Dataset Generation with Few-shot Guidance
Per-image LoRA adapters fused at inference time produce synthetic training data that improves few-shot image classification accuracy over existing synthetic-data methods.
-
Does Feasibility Matter? Understanding the Impact of Feasibility on Synthetic Training Data
Feasibility of synthetic images has little effect on fine-tuned CLIP accuracy; the edited attribute (background, color, or texture) matters more than whether the attribute is realistic.
-
AI-Generated Lecture Slides for Improving Slide Element Detection and Retrieval
Synthetic lecture slides generated by an LLM pipeline improve few-shot slide element detection and text-based retrieval when used as pre-training data.
Discussion (0). Continue with ORCID to comment.