mlsanityData to 14 Oct 2026

S3OD: Towards Generalizable Salient Object Detection with Synthetic Data

arXiv 2510.21605 ↗First posted 24 Oct 2025cs.CVcited by 6

Not on the map. Why?

Abstract (arXiv)

Salient object detection exemplifies data-bounded tasks where expensive pixel-precise annotations force separate model training for related subtasks like DIS and HR-SOD. We present a method that dramatically improves generalization through large-scale synthetic data generation and ambiguity-aware architecture. We introduce S3OD, a dataset of over 139,000 high-resolution images created through our multi-modal diffusion pipeline that extracts labels from diffusion and DINO-v3 features. The iterative generation framework prioritizes challenging categories based on model performance. We propose a streamlined multi-mask decoder that handles the inherent ambiguity in salient object detection by predicting multiple valid interpretations. Models trained only on synthetic data achieve 20-50% error reduction in cross-dataset generalization, while fine-tuned versions reach state-of-the-art performance across DIS and HR-SOD benchmarks.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 49 of its 76 references that are in the map. First posted on arXiv 24 Oct 2025.

Builds on

49 of 76 references.

References · 49

  1. 2017Attention Is All You Needcited by 60,361
  2. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  4. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  5. 2023GPT-4 Technical Reportcited by 18,396
  6. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  7. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,152
  8. 2018UMAP: Uniform Manifold Approximation and Projection for Dimension Reductioncited by 2,898
  9. 2024SAM 2: Segment Anything in Images and Videoscited by 2,863
  10. 2025Gemma 3 Technical Reportcited by 1,824
  11. 2021Vision Transformers for Dense Predictioncited by 1,803
  12. 2025DINOv3cited by 1,414
37 more
  1. 2017The 2017 DAVIS Challenge on Video Object Segmentationcited by 957
  2. 2019Improved Precision and Recall Metric for Assessing Generative Modelscited by 944
  3. 2017Structure-measure: A New Way to Evaluate Foreground Mapscited by 480
  4. 2020U2-Net: Going Deeper with Nested U-Structure for Salient Object Detectioncited by 412
  5. 2023Segment Anything in High Qualitycited by 376
  6. 2018Enhanced-alignment Measure for Binary Foreground Map Evaluationcited by 370
  7. 2023Synthetic Data from Diffusion Models Improves ImageNet Classificationcited by 327
  8. 2019EGNet:Edge Guidance Network for Salient Object Detectioncited by 237
  9. 2021Anabranch Network for Camouflaged Object Segmentationcited by 227
  10. 2019Cascaded Partial Decoder for Fast and Accurate Salient Object Detectioncited by 216
  11. 2021Simultaneously Localize, Segment and Rank the Camouflaged Objectscited by 198
  12. 2023StableRep: Synthetic Images from Text-to-Image Models Make Strong Visual Representation Learnerscited by 177
  13. 2022Fake it till you make it: Learning transferable representations from synthetic ImageNet clonescited by 172
  14. 2015Salient Object Detection: A Benchmarkcited by 131
  15. 2023DatasetDM: Synthesizing Data with Perception Annotations Using Diffusion Modelscited by 123
  16. 2024Bilateral Reference for High-Resolution Dichotomous Image Segmentationcited by 112
  17. 2020Label Decoupling Framework for Salient Object Detectioncited by 111
  18. 2022Highly Accurate Dichotomous Image Segmentationcited by 110
  19. 2023Scaling Laws of Synthetic Images for Model Training ... for Nowcited by 108
  20. 2016Learning in an Uncertain World: Representing Ambiguity Through Multiple Hypothesescited by 97
  21. 2019Towards High-Resolution Salient Object Detectioncited by 92
  22. 2023Feature Shrinkage Pyramid for Camouflaged Object Detection with Transformerscited by 87
  23. 2023Diversity is Definitely Needed: Improving Model-Agnostic Zero-shot Classification via Stable Diffusioncited by 81
  24. 2023Diffusion Models for Open-Vocabulary Segmentationcited by 64
  25. 2022Pyramid Grafting Network for One-Stage High Resolution Saliency Detectioncited by 63
  26. 2025ConceptAttention: Diffusion Transformers Learn Highly Interpretable Featurescited by 52
  27. 2022Revisiting Image Pyramid Structure for High Resolution Salient Object Detectioncited by 36
  28. 2021Disentangled High Quality Salient Object Detectioncited by 35
  29. 2024Multi-view Aggregation Network for Dichotomous Image Segmentationcited by 28
  30. 2024Dataset Enhancement with Instance-Level Augmentationscited by 23
  31. 2023Unite-Divide-Unite: Joint Boosting Trunk and Structure for High-accuracy Dichotomous Image Segmentationcited by 17
  32. 2021Fully Understanding Generic Objects: Modeling, Segmentation, and Reconstructioncited by 16
  33. 2022Synthetic Data Supervised Salient Object Detectioncited by 15
  34. 2024Mask Factory: Towards High-quality Synthetic Data Generation for Dichotomous Image Segmentationcited by 15
  35. 2023Exploring Limits of Diffusion-Synthetic Training with Weakly Supervised Semantic Segmentationcited by 4
  36. 2023Promoting Segment Anything Model towards Highly Accurate Dichotomous Image Segmentationcited by 4
  37. 2024VGGHeads: 3D Multi Head Alignment with a Large-Scale Synthetic Datasetcited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 5.

6 papers.

  1. 2025RealCamo: Boosting Real Camouflage Synthesis with Layout Controls and Textual-Visual Guidancecited by 1
  2. 2026Beyond Pixel Overlap: A Framework for Decomposing Segmentation Evaluation Metricscited by 1
  3. 2026PoseDreamer: Scalable and Photorealistic Human Data Generation Pipeline with Diffusion Modelscited by 0
  4. 2026MMDiff: Extending Diffusion Transformers for Multi-Modal Generationcited by 0
  5. 2026FLAT: Feedforward Latent Triangle Splatting for Geometrically Accurate Scene Generationcited by 0
  6. 2026Level-of-Token Diffusioncited by 0