mlsanityData to 14 Oct 2026

StableRep: Synthetic Images from Text-to-Image Models Make Strong Visual Representation Learners

arXiv 2306.00984 ↗First posted 1 Jun 2023cs.CVcited by 177

Not on the map. Why?

Cited by 38 in Diffusion models, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

We investigate the potential of learning visual representations using synthetic images generated by text-to-image models. This is a natural question in the light of the excellent performance of such models in generating high-quality images. We consider specifically the Stable Diffusion, one of the leading open source text-to-image models. We show that (1) when the generative model is configured with proper classifier-free guidance scale, training self-supervised methods on synthetic images can match or beat the real image counterpart; (2) by treating the multiple images generated from the same text prompt as positives for each other, we develop a multi-positive contrastive learning method, which we call StableRep. With solely synthetic images, the representations learned by StableRep surpass the performance of representations learned by SimCLR and CLIP using the same set of text prompts and corresponding real images, on large scale datasets. When we further add language supervision, StableRep trained with 20M synthetic images achieves better accuracy than CLIP trained with 50M real images.

Key facts. Cited by 177 papers in the map (data through 14 Oct 2026). Builds on 67 of its 93 references that are in the map. First posted on arXiv 1 Jun 2023.

Builds on

67 of 93 references.

References · 67

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  4. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  5. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 12,049
  6. 2020Denoising Diffusion Implicit Modelscited by 9,781
  7. 2018Representation Learning with Contrastive Predictive Codingcited by 7,777
  8. 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,513
  9. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,403
  10. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
  11. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 7,005
  12. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,968
38 more
  1. 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 6,585
  2. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 6,355
  3. 2022Classifier-Free Diffusion Guidancecited by 5,847
  4. 2023Adding Conditional Control to Text-to-Image Diffusion Modelscited by 5,503
  5. 2020Bootstrap your own latent: A new approach to self-supervised Learningcited by 4,750
  6. 2022DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generationcited by 3,300
  7. 2020Supervised Contrastive Learningcited by 2,829
  8. 2013Describing Textures in the Wildcited by 2,125
  9. 2018Unsupervised Feature Learning via Non-Parametric Instance-level Discriminationcited by 1,845
  10. 2013Fine-Grained Visual Classification of Aircraftcited by 1,726
  11. 2021An Empirical Study of Training Self-Supervised Vision Transformerscited by 1,616
  12. 2021LAION-400M: Open Dataset of CLIP-Filtered 400 Million Image-Text Pairscited by 1,555
  13. 2016Context Encoders: Feature Learning by Inpaintingcited by 1,428
  14. 2019Contrastive Multiview Codingcited by 1,301
  15. 2022Scaling Autoregressive Models for Content-Rich Text-to-Image Generationcited by 1,295
  16. 2018Unified Perceptual Parsing for Scene Understandingcited by 1,271
  17. 2021Conceptual 12M: Pushing Web-Scale Image-Text Pre-Training To Recognize Long-Tail Visual Conceptscited by 1,216
  18. 2022Imagic: Text-Based Real Image Editing with Diffusion Modelscited by 1,178
  19. 2022MaskGIT: Masked Generative Image Transformercited by 1,099
  20. 2016Semantic Understanding of Scenes through the ADE20K Datasetcited by 1,069
  21. 2015FlowNet: Learning Optical Flow with Convolutional Networkscited by 1,059
  22. 2022eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoiserscited by 931
  23. 2015A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimationcited by 881
  24. 2020What Makes for Good Views for Contrastive Learning?cited by 846
  25. 2023Muse: Text-To-Image Generation via Masked Generative Transformerscited by 648
  26. 2022When and why vision-language models behave like bags-of-words, and what to do about it?cited by 610
  27. 2018Do Better ImageNet Models Transfer Better?cited by 565
  28. 2023Scaling up GANs for Text-to-Image Synthesiscited by 492
  29. 2020Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need?cited by 437
  30. 2021SLIP: Self-supervision meets Language-Image Pre-trainingcited by 436
  31. 2017Learning from Synthetic Humanscited by 359
  32. 2023Synthetic Data from Diffusion Models Improves ImageNet Classificationcited by 327
  33. 2022Is synthetic data from generative models ready for image recognition?cited by 320
  34. 2019A Baseline for Few-Shot Image Classificationcited by 298
  35. 2020How Well Do Self-Supervised Models Transfer?cited by 213
  36. 2022Generating Training Data with Language Models: Towards Zero-Shot Language Understandingcited by 199
  37. 2019SimpleShot: Revisiting Nearest-Neighbor Classification for Few-Shot Learningcited by 195
  38. 2021RedCaps: web-curated image-text data created by the people, for the peoplecited by 180

50 most cited of 67 All 67 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 25; 2024: 77; 2025: 51; 2026: 24.

177 papers.

Sort:

By area:

  1. 2023JourneyDB: A Benchmark for Generative Image Understandingcited by 205
  2. 2024Eyes Closed, Safety On: Protecting Multimodal LLMs via Image-to-Text Transformationcited by 124
  3. 2024URDFormer: A Pipeline for Constructing Articulated Simulation Environments from Real-World Imagescited by 111
  4. 2023Scaling Laws of Synthetic Images for Model Training ... for Nowcited by 108
  5. 2024No "Zero-Shot" Without Exponential Data: Pretraining Concept Frequency Determines Multimodal Model Performancecited by 75
  6. 2024DreamLIP: Language-Image Pre-training with Long Captionscited by 72
  7. 2025Vision-Language Models Do Not Understand Negationcited by 72
  8. 2023Learning Vision from Models Rivals Learning Vision from Datacited by 71
  9. 2023Generative AI for Medical Imaging: extending the MONAI Frameworkcited by 68
  10. 2024SynthCLIP: Are We Ready for a Fully Synthetic CLIP Training?cited by 64
  11. 2023NoiseCLR: A Contrastive Learning Approach for Unsupervised Discovery of Interpretable Directions in Diffusion Modelscited by 47
  12. 2025Enhanced Contrastive Learning with Multi-view Longitudinal Data for Chest X-ray Report Generationcited by 42
38 more
  1. 2024Is Synthetic Data all We Need? Benchmarking the Robustness of Models Trained with Synthetic Imagescited by 41
  2. 2024Diffusion Feedback Helps CLIP See Bettercited by 38
  3. 2024TIPS: Text-Image Pretraining with Spatial awarenesscited by 37
  4. 2023Real-Fake: Effective Training Data Synthesis Through Distribution Matchingcited by 36
  5. 2024A Vision Check-up for Language Modelscited by 36
  6. 2024Consistency-diversity-realism Pareto fronts of conditional image generative modelscited by 36
  7. 2023Gaining Wisdom from Setbacks: Aligning Large Language Models via Mistake Analysiscited by 35
  8. 2024Enhance Image Classification via Inter-Class Image Mixup with Diffusion Modelcited by 34
  9. 2023Feedback-guided Data Synthesis for Imbalanced Classificationcited by 33
  10. 2024DataDream: Few-shot Guided Dataset Generationcited by 29
  11. 2024Learning General-Purpose Biomedical Volume Representations using Randomized Synthesiscited by 29
  12. 2024LoTLIP: Improving Language-Image Pre-training for Long Text Understandingcited by 27
  13. 2024Do Generated Data Always Help Contrastive Learning?cited by 25
  14. 2024Would Deep Generative Models Amplify Bias in Future Models?cited by 24
  15. 2024Dataset Enhancement with Instance-Level Augmentationscited by 23
  16. 2024ImageNet-D: Benchmarking Neural Network Robustness on Diffusion Synthetic Objectcited by 22
  17. 2024DKDM: Data-Free Knowledge Distillation for Diffusion Models with Any Architecturecited by 22
  18. 2025Synthetic Data is an Elegant GIFT for Continual Vision-Language Modelscited by 21
  19. 2024Bootstrapping Language-Guided Navigation Learning with Self-Refining Data Flywheelcited by 20
  20. 2023ConvNet vs Transformer, Supervised vs CLIP: Beyond ImageNet Accuracycited by 19
  21. 2023Mitigating Open-Vocabulary Caption Hallucinationscited by 19
  22. 2024The Unmet Promise of Synthetic Training Images: Using Retrieved Real Images Performs Bettercited by 19
  23. 2024Minority-Focused Text-to-Image Generation via Prompt Optimizationcited by 18
  24. 2025Improving the Scaling Laws of Synthetic Data with Deliberate Practicecited by 17
  25. 2023HiPA: Enabling One-Step Text-to-Image Diffusion Models via High-Frequency-Promoting Adaptationcited by 16
  26. 2025GenHancer: Imperfect Generative Models are Secretly Strong Vision-Centric Enhancerscited by 16
  27. 2024SELMA: Learning and Merging Skill-Specific Text-to-Image Experts with Auto-Generated Datacited by 14
  28. 2024GLASS: Guided Latent Slot Diffusion for Object-Centric Learningcited by 14
  29. 2025A Large Scale Analysis of Gender Biases in Text-to-Image Generative Modelscited by 14
  30. 2023AI-Generated Images as Data Source: The Dawn of Synthetic Eracited by 13
  31. 2023Gen2Det: Generate to Detectcited by 12
  32. 2023Learning Subject-Aware Cropping by Outpainting Professional Photoscited by 12
  33. 2024FLAME: Frozen Large Language Models Enable Data-Efficient Language-Image Pre-trainingcited by 12
  34. 2024Detecting Human Artifacts from Text-to-Image Modelscited by 12
  35. 2025Enhancing Vision-Language Compositional Understanding with Multimodal Synthetic Datacited by 12
  36. 2023Diversified in-domain synthesis with efficient fine-tuning for few-shot classificationcited by 11
  37. 2025TULIP: Towards Unified Language-Image Pretrainingcited by 11
  38. 2024MAGID: An Automated Pipeline for Generating Synthetic Multi-modal Datasetscited by 10

50 most cited of 177 All 177 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2023Diversity and Diffusion: Observations on Synthetic Image Distributions with Stable Diffusion
  2. 2022Evaluating a Synthetic Image Dataset Generated with Stable Diffusion
  3. 2024FakeInversion: Learning to Detect Images from Unseen Text-to-Image Models by Inverting Stable Diffusion
  4. 2025NSYNC: Negative Synthetic Image Generation for Contrastive Training to Improve Stylized Text-To-Image Translation
  5. 2023Diversify, Don't Fine-Tune: Scaling Up Visual Recognition Training with Synthetic Images
  6. 2023Text-to-Image Diffusion Models are Zero-Shot Classifiers
  7. 2023ALIP: Adaptive Language-Image Pre-training with Synthetic Caption
  8. 2024Synth2: Boosting Visual-Language Models with Synthetic Captions and Image Embeddings