S3OD: Towards Generalizable Salient Object Detection with Synthetic Data
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917304569167872 |
|---|---|
| author | Kupyn, Orest Kataoka, Hirokatsu Rupprecht, Christian |
| author_facet | Kupyn, Orest Kataoka, Hirokatsu Rupprecht, Christian |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_21605 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | S3OD: Towards Generalizable Salient Object Detection with Synthetic Data Kupyn, Orest Kataoka, Hirokatsu Rupprecht, Christian Computer Vision and Pattern Recognition 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. |
| title | S3OD: Towards Generalizable Salient Object Detection with Synthetic Data |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2510.21605 |