Research map / Detection and segmentation / Open-set and few-shot detection
Texture-Guided Saliency Distilling for Unsupervised Salient Object Detection
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 CSD, which gradually increases the training emphasis on low-confidence pixels (ablation) | Texture-guided saliency distilling · RGB SOD pseudo-label quality | Fβ: .908 -> .917 | Helped |
| Add BTM to CSD and multi-scale consistency (ablation) | Texture-guided saliency distilling · RGB SOD pseudo-label quality | Fβ: .908 -> .917 | Helped |
| Add multi-scale consistency loss to L_adb (ablation) | Texture-guided saliency distilling · RGB SOD pseudo-label quality | Fβ: .882 -> .891 | Helped |
| Replace CSD with BCE loss (ablation) | Texture-guided saliency distilling · RGB SOD pseudo-label quality | Fβ: .917 -> .558 | Hurt |
| Replace CSD with L1 loss (ablation) | Texture-guided saliency distilling · RGB SOD pseudo-label quality | Fβ: .917 -> .895 | Hurt |
| Replace CSD with L_adb (ablation) | Texture-guided saliency distilling · RGB SOD pseudo-label quality | Fβ: .917 -> .903 | Hurt |
| Replace CSD with L_spl (ablation) | Texture-guided saliency distilling · RGB SOD pseudo-label quality | Fβ: .917 -> .910 | Hurt |
| Use L1 distance for appearance-space texture features in BTM (ablation) | Texture-guided saliency distilling · RGB SOD pseudo-label quality | Fβ: .917 -> .735 | Hurt |
| Replace BTM with L_lsc (ablation) | Texture-guided saliency distilling · RGB SOD pseudo-label quality | Fβ: .917 -> .906 | Hurt |
| Remove boundary masks from BTM (ablation) | Texture-guided saliency distilling · RGB SOD pseudo-label quality | Fβ: .917 -> .907 | Hurt |
| Use multimodal training data instead of task-specific data (ablation) | Texture-guided saliency distilling · RGB-D SOD | Fβ: .834 -> .877 | Helped |
| Use multimodal training data instead of task-specific data (ablation) | Texture-guided saliency distilling · Video SOD | Fβ: .534 -> .547 (DAVSOD); .751 -> .808 (SegV2); .732 -> .795 (FBMS); .751 -> .756 (DAVIS) | Mixed |
Closest papers by meaning
- Promoting Saliency From Depth: Deep Unsupervised RGB-D Saliency Detection (ICLR 2022 · similarity 0.91)
- Unsupervised Domain Adaptive Salient Object Detection through Uncertainty-Aware Pseudo-Label Learning (AAAI 2022 · similarity 0.90)
- DeepUSPS: Deep Robust Unsupervised Saliency Prediction via Self-supervision (NeurIPS 2019 · similarity 0.90)
- Deep Unsupervised Saliency Detection: A Multiple Noisy Labeling Perspective (CVPR 2018 · similarity 0.89)
- BASNet: Boundary-Aware Salient Object Detection (CVPR 2019 · similarity 0.89)
- Unsupervised Salient Instance Detection (CVPR 2024 · similarity 0.88)
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.