Research map / Vision-language and multimodal / Human and embodied modeling
Understanding and Visualizing Deep Visual Saliency Models
Check 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 |
|---|---|---|---|
| Use deep saliency models rather than classical saliency models | saliency prediction on MIT300 and synthetic search arrays · MIT300; Synthetic search arrays | NSS; NMM: DeepGaze II 2.34 NSS; classical BMS 3.65 NMM | Mixed |
| Fine-tune VGG-16 for saliency prediction (ablation) | pre-trained VGG-16 saliency model · OSIE-SR, person head | mean NSS for top 10 activation maps: 1.21 -> 3.38 | Helped |
| Fine-tune VGG-16 for saliency prediction (ablation) | pre-trained VGG-16 saliency model · OSIE-SR, text | mean NSS for top 10 activation maps: 1.1 -> 1.38 | Helped |
| Fine-tune more than 3 VGG-16 layers (ablation) | saliency prediction model · OSIE-SR salient categories | mean NSS and number of activation maps | Mixed |
| Fine-tune VGG-16 for saliency prediction (ablation) | pre-trained VGG-16 · Synthetic search arrays | NMM: 0.38 -> 0.8; 0.98; 0.92; 0.65; 0.86; 0.83; 0.73 | Hurt |
| Fine-tune conv5-3 for saliency prediction instead of scene recognition (ablation) | VGG-16 trained on the same SALICON-SAL-SCE images · SALICON-SAL-SCE salient categories | mean NSS for top 10 activation maps: pt 1.21; sp 2.51; sr 0.47 | Helped |
| Fine-tune VGG-16 for scene recognition (ablation) | pre-trained VGG-16 · SALICON-SAL-SCE salient categories | mean NSS for top 10 activation maps: pt 1.21; sr 0.47 | Hurt |
| Fine-tune more VGG-16 layers (ablation) | saliency prediction model · Synthetic search arrays, conv5-3 activation maps | NMM: 0 layers: 0.38; 1: 0.8; 2: 0.98; 3: 0.92; 4: 0.65; 5: 0.86; 6: 0.83; all: 0.73 | Mixed |
Closest papers by meaning
- A Deeper Look at Saliency: Feature Contrast, Semantics, and Beyond (CVPR 2016 · similarity 0.93)
- What Do Deep Saliency Models Learn about Visual Attention? (NeurIPS 2023 · similarity 0.93)
- Shallow and Deep Convolutional Networks for Saliency Prediction (CVPR 2016 · similarity 0.92)
- Predicting When Saliency Maps Are Accurate and Eye Fixations Consistent (CVPR 2016 · similarity 0.90)
- There and Back Again: Revisiting Backpropagation Saliency Methods (CVPR 2020 · similarity 0.89)
- CaRDiff: Video Salient Object Ranking Chain of Thought Reasoning for Saliency Prediction with Diffusion (AAAI 2025 · similarity 0.89)
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.