mlsanityData to 7 Oct 2026

Contrastive Syn-to-Real Generalization

arXiv 2104.02290 ↗First posted 6 Apr 2021cs.CVcited by 32

Not on the map. Why?

Abstract (arXiv)

Training on synthetic data can be beneficial for label or data-scarce scenarios. However, synthetically trained models often suffer from poor generalization in real domains due to domain gaps. In this work, we make a key observation that the diversity of the learned feature embeddings plays an important role in the generalization performance. To this end, we propose contrastive synthetic-to-real generalization (CSG), a novel framework that leverages the pre-trained ImageNet knowledge to prevent overfitting to the synthetic domain, while promoting the diversity of feature embeddings as an inductive bias to improve generalization. In addition, we enhance the proposed CSG framework with attentional pooling (A-pool) to let the model focus on semantically important regions and further improve its generalization. We demonstrate the effectiveness of CSG on various synthetic training tasks, exhibiting state-of-the-art performance on zero-shot domain generalization.

Key facts. Cited by 32 papers in the map (data through 7 Oct 2026). Builds on 38 of its 42 references that are in the map. First posted on arXiv 6 Apr 2021.

Builds on

38 of 42 references.

References · 38

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  3. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
  4. 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
  5. 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,482
  6. 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,690
  7. 2016Overcoming catastrophic forgetting in neural networkscited by 4,718
  8. 2016The Cityscapes Dataset for Semantic Urban Scene Understandingcited by 4,427
  9. 2016Learning without Forgettingcited by 2,465
  10. 2015Learning Deep Features for Discriminative Localizationcited by 2,430
  11. 2020Improved Baselines with Momentum Contrastive Learningcited by 2,282
  12. 2019RandAugment: Practical automated data augmentation with a reduced search spacecited by 2,261
26 more
  1. 2017Rethinking Atrous Convolution for Semantic Image Segmentationcited by 2,154
  2. 2018Unsupervised Feature Learning via Non-Parametric Instance-level Discriminationcited by 1,838
  3. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,430
  4. 2020Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hyperspherecited by 1,393
  5. 2020Big Self-Supervised Models are Strong Semi-Supervised Learnerscited by 1,375
  6. 2019Contrastive Multiview Codingcited by 1,297
  7. 2018Learning deep representations by mutual information estimation and maximizationcited by 1,265
  8. 2019Habitat: A Platform for Embodied AI Researchcited by 1,163
  9. 2017Deeper, Broader and Artier Domain Generalizationcited by 983
  10. 2020What Makes for Good Views for Contrastive Learning?cited by 843
  11. 2016Playing for Data: Ground Truth from Computer Gamescited by 816
  12. 2019Self-Supervised Learning of Pretext-Invariant Representationscited by 806
  13. 2017Learning to Generalize: Meta-Learning for Domain Generalizationcited by 762
  14. 2019Contrastive Representation Distillationcited by 734
  15. 2019Data-Efficient Image Recognition with Contrastive Predictive Codingcited by 672
  16. 2013Domain Generalization via Invariant Feature Representationcited by 622
  17. 2017VisDA: The Visual Domain Adaptation Challengecited by 432
  18. 2018Two at Once: Enhancing Learning and Generalization Capacities via IBN-Netcited by 361
  19. 2016Learning from Simulated and Unsupervised Images through Adversarial Trainingcited by 287
  20. 2019Domain Randomization and Pyramid Consistency: Simulation-to-Real Generalization without Accessing Target Domain Datacited by 245
  21. 2020Robust Pre-Training by Adversarial Contrastive Learningcited by 144
  22. 2017ROAD: Reality Oriented Adaptation for Semantic Segmentation of Urban Scenescited by 95
  23. 2018Learning towards Minimum Hyperspherical Energycited by 75
  24. 2018Self-produced Guidance for Weakly-supervised Object Localizationcited by 72
  25. 2016Learning Attributes Equals Multi-Source Domain Generalizationcited by 44
  26. 2020Automated Synthetic-to-Real Generalizationcited by 35

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 5; 2022: 13; 2023: 9; 2024: 3; 2025: 0; 2026: 2.

32 papers.

Sort:

By area:

  1. 2024SAT: Dynamic Spatial Aptitude Training for Multimodal Language Modelscited by 111
  2. 2023Label-Free Liver Tumor Segmentationcited by 56
  3. 2022Siamese Contrastive Embedding Network for Compositional Zero-Shot Learningcited by 55
  4. 2022Pin the Memory: Learning to Generalize Semantic Segmentationcited by 48
  5. 2021Source-Free Open Compound Domain Adaptation in Semantic Segmentationcited by 43
  6. 2022PASTA: Proportional Amplitude Spectrum Training Augmentation for Syn-to-Real Domain Generalizationcited by 38
  7. 2022Behavioral Intention Prediction in Driving Scenes: A Surveycited by 30
  8. 2023Texture Learning Domain Randomization for Domain Generalized Segmentationcited by 23
  9. 2022Few-Shot Adaptation of Pre-Trained Networks for Domain Shiftcited by 16
  10. 2024Source-Free Domain Adaptation Guided by Vision and Vision-Language Pre-Trainingcited by 15
  11. 2021A Scaling Law for Synthetic-to-Real Transfer: How Much Is Your Pre-training Effective?cited by 14
  12. 2022Rethinking the Role of Pre-Trained Networks in Source-Free Domain Adaptationcited by 13
20 more
  1. 2023Generalized Semantic Segmentation by Self-Supervised Source Domain Projection and Multi-Level Contrastive Learningcited by 11
  2. 2023A New Benchmark: On the Utility of Synthetic Data with Blender for Bare Supervised Learning and Downstream Domain Adaptationcited by 11
  3. 2023Pick the Best Pre-trained Model: Towards Transferability Estimation for Medical Image Segmentationcited by 10
  4. 2021Self-Supervised Learning of Domain Invariant Features for Depth Estimationcited by 9
  5. 2021Adversarial Semantic Hallucination for Domain Generalized Semantic Segmentationcited by 9
  6. 2022Semantic Self-adaptation: Enhancing Generalization with a Single Samplecited by 8
  7. 2023SUG: Single-dataset Unified Generalization for 3D Point Cloud Classificationcited by 6
  8. 2023BELT:Bootstrapping Electroencephalography-to-Language Decoding and Zero-Shot Sentiment Classification by Natural Language Supervisioncited by 5
  9. 2022Reducing the Amount of Real World Data for Object Detector Training with Synthetic Datacited by 4
  10. 2022TRoVE: Transforming Road Scene Datasets into Photorealistic Virtual Environmentscited by 2
  11. 2022Feature diversity in self-supervised learningcited by 2
  12. 2023Mind the Backbone: Minimizing Backbone Distortion for Robust Object Detectioncited by 2
  13. 2026Learning Data-Efficient and Generalizable Neural Operators via Fundamental Physics Knowledgecited by 2
  14. 2022GCISG: Guided Causal Invariant Learning for Improved Syn-to-real Generalizationcited by 1
  15. 2024Adversarial training of Keyword Spotting to Minimize TTS Data Overfittingcited by 1
  16. 2021P-WAE: Generalized Patch-Wasserstein Autoencoder for Anomaly Screeningcited by 0
  17. 2022Few-Max: Few-Shot Domain Adaptation for Unsupervised Contrastive Representation Learningcited by 0
  18. 2022Generalization Beyond Feature Alignment: Concept Activation-Guided Contrastive Learningcited by 0
  19. 2023SelectNAdapt: Support Set Selection for Few-Shot Domain Adaptationcited by 0
  20. 2026Latent space mapping of interpretable structural coordinates from stochastic single-molecule signalscited by 0