Research map / Detection and segmentation / CNN features
S3OD: Towards Generalizable Salient Object Detection with Synthetic Data
Orest Kupyn, Hirokatsu Kataoka, Christian Rupprecht
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add iterative hard-sample data generation (3 rounds vs single round) (ablation) | S3OD data · DIS and DUT-OMRON | F-measure: DUT-OMRON .855 -> .860 | Improved |
| add DINO-v3 features to data diffusion (ablation) | S3OD labeling · DUT-OMRON | F-measure: .681 -> .873 | Improved |
| add DiT maps to data diffusion (ablation) | S3OD labeling · DUT-OMRON | F-measure: .870 -> .873 | Improved |
| add concept maps to data diffusion (ablation) | S3OD labeling · DUT-OMRON | F-measure: .871 -> .873 | Improved |
| replace Swin-B with DINO-v3 backbone (ablation) | S3ODNet · DIS-5K | F-measure: .884 -> .909 | Improved |
| use three predicted masks instead of one (ablation) | S3ODNet · DIS-5K | F-measure: .909 -> .914 | Improved |
| use two predicted masks instead of one (ablation) | S3ODNet · DIS-5K | F-measure: .909 -> .892 | Hurt |
| use LLM-generated prompts instead of class names (ablation) | S3OD generation · synthetic image quality | Inception Score: 67.8 -> 98.1 | Improved |
| train on S3OD synthetic data only instead of DIS-5K | S3ODNet · DUT-OMRON cross-dataset | F-measure: .808 -> .860 | Improved |
| retrain BiRefNet on S3OD instead of DIS-5K | BiRefNet · DUT-OMRON | F-measure: .744 -> .845 | Improved |
| add MaskFactory synthetic data to DIS-5K (ablation) | S3ODNet · DUT-OMRON | F-measure: .808 -> .774 | Hurt |
| add DatasetDM synthetic data to DIS-5K (ablation) | S3ODNet · DUT-OMRON | F-measure: .808 -> .770 | Hurt |
Closest papers by meaning
- DiMSOD: A Diffusion-Based Framework for Multi-Modal Salient Object Detection (AAAI 2025 · similarity 0.88)
- DVSOD: RGB-D Video Salient Object Detection (NeurIPS 2023 · similarity 0.88)
- SimpleDiffusion: A Lightweight and Efficient Conditional Diffusion Model for Multi-Modal Salient Object Detection (AAAI 2026 · similarity 0.88)
- Breaking Barriers, Localizing Saliency: A Large-Scale Benchmark and Baseline for Condition-Constrained Salient Object Detection (TPAMI 2025 · similarity 0.88)
- Unsupervised Domain Adaptive Salient Object Detection through Uncertainty-Aware Pseudo-Label Learning (AAAI 2022 · similarity 0.88)
- Generalized Diffusion Detector: Mining Robust Features from Diffusion Models for Domain-Generalized Detection (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
