Research map / Architectures and efficiency / Attention and transformers
Visual Saliency Transformer
Nian Liu, Ni Zhang, Kaiyuan Wan, Ling Shao, Junwei Han
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 cross-modality transformer fusion instead of concatenation (ablation) | RGB-D VST baseline · LFSD | S-measure: 0.842 -> 0.849 | Improved |
| add cross-modality transformer fusion instead of concatenation (ablation) | RGB-D VST baseline · DUTLF-Depth | S-measure: 0.889 -> 0.889 | No clear effect |
| add bilinear upsampling to full resolution (ablation) | +CMT · NJUD | S-measure: 0.873 -> 0.906 | Improved |
| replace bilinear upsampling with RT2T token upsampling (ablation) | +CMT+Bili · NJUD | S-measure: 0.906 -> 0.915 | Improved |
| replace bilinear upsampling with RT2T token upsampling (ablation) | +RC+Bili · DUTS | S-measure: 0.867 -> 0.881 | Improved |
| add multi-level token fusion of low-level tokens (ablation) | +CMT+RT2T · STERE | S-measure: 0.896 -> 0.910 | Improved |
| add multi-level token fusion of low-level tokens (ablation) | +RC+RT2T · DUTS | S-measure: 0.881 -> 0.895 | Improved |
| add token-based multi-task decoder with boundary detection (ablation) | +CMT+RT2T+F · NJUD | S-measure: 0.923 -> 0.922 | No clear effect |
| add token-based multi-task decoder with boundary detection (ablation) | +CMT+RT2T+F · LFSD | S-measure: 0.876 -> 0.882 | Improved |
| add token-based multi-task decoder with boundary detection (ablation) | +RC+RT2T+F · DUTS | S-measure: 0.895 -> 0.896 | Improved |
| use task tokens decoder instead of conventional two-stream decoder (ablation) | +CMT+RT2T+F · LFSD | S-measure: 0.874 -> 0.882 | Improved |
| use task tokens decoder instead of conventional two-stream decoder (ablation) | +CMT+RT2T+F · MACs | MACs (G): 28.27 -> 17.70 | Improved |
Closest papers by meaning
- VST++: Efficient and Stronger Visual Saliency Transformer (TPAMI 2024 · similarity 0.96)
- Saliency-Driven Token Merging for Vision Transformers (CVPR 2026 · similarity 0.90)
- Promoting Saliency From Depth: Deep Unsupervised RGB-D Saliency Detection (ICLR 2022 · similarity 0.89)
- Learning Generative Vision Transformer with Energy-Based Latent Space for Saliency Prediction (NeurIPS 2021 · similarity 0.89)
- DVSOD: RGB-D Video Salient Object Detection (NeurIPS 2023 · similarity 0.88)
- Alignment-Free RGB-T Salient Object Detection: A Large-Scale Dataset and Progressive Correlation Network (AAAI 2025 · 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.
