Research map / Detection and segmentation / CNN features
Task-driven Webpage Saliency
Quanlong Zheng, Jianbo Jiao, Ying Cao, Rynson W.H. Lau
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 task-specific subnet (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | NSS: 0.412 -> 0.622 | Improved |
| add task-specific subnet (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | KL: 1.330 -> 0.883 | Improved |
| add task-free subnet (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | NSS: 0.559 -> 0.622 | Improved |
| add task-free subnet (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | KL: 0.810 -> 0.883 | Hurt |
| share encoder between segmentation and task-free subnets instead of separate encoders (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | NSS: 0.566 -> 0.622 | Improved |
| share encoder between segmentation and task-free subnets instead of separate encoders (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | KL: 1.013 -> 0.883 | Improved |
| use unified model with task input instead of 5 separate per-task CNNs (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | NSS: 0.498 -> 0.622 | Improved |
| pre-train task-specific subnet on synthetic task-driven data (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | KL: 10.428 -> 0.883 | Improved |
| pre-train task-specific subnet on synthetic task-driven data (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | NSS: 0.337 -> 0.622 | Improved |
| add fine-tuning on real data after synthetic data (vs synthetic only) (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | NSS: 0.552 -> 0.622 | Improved |
| add fine-tuning on real data after synthetic data (vs synthetic only) (ablation) | Task-driven saliency network · Task-driven webpage saliency (200 webpages) | KL: 2.722 -> 0.883 | Improved |
Closest papers by meaning
- Learning From Unique Perspectives: User-Aware Saliency Modeling (CVPR 2023 · similarity 0.87)
- What Do Deep Saliency Models Learn about Visual Attention? (NeurIPS 2023 · similarity 0.87)
- Does Text Attract Attention on E-Commerce Images: A Novel Saliency Prediction Dataset and Method (CVPR 2022 · similarity 0.86)
- Understanding and Visualizing Deep Visual Saliency Models (CVPR 2019 · similarity 0.86)
- Personalized Saliency and Its Prediction (TPAMI 2018 · similarity 0.86)
- CaRDiff: Video Salient Object Ranking Chain of Thought Reasoning for Saliency Prediction with Diffusion (AAAI 2025 · similarity 0.85)
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.
