Research map / Video / Video-language understanding
Saliency Prediction in the Deep Learning Era: Successes and Limitations
Ali Borji
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 |
|---|---|---|---|
| use deep (CNN) saliency models instead of best classic model | saliency models · MIT300 | NSS: 1.41 -> 2.47 | Improved |
| use deep models instead of best classic model | saliency models · MIT300 | AUC-Judd: 0.83 -> 0.88 | Improved |
| use deep models instead of best non-deep model | saliency models · CAT2000 | AUC-Judd: 0.85 -> 0.88 | Improved |
| use deep models instead of best non-deep model | saliency models · CAT2000 | NSS: 1.67 -> 2.39 | Improved |
| switch evaluation metric from AUC-Judd to AUC-Borji | EML-NET · MIT300 | rank: 2nd -> 44th | Mixed |
| add temporal information to classic video saliency models vs adding deep CNNs | video saliency models · DHF1K | NSS/CC/AUC | Improved |
| use a subset of 10 representative images to rank models | MIT benchmark · MIT300 | Spearman correlation: 0.97 | Improved |
| replace fixations with mouse data from 90 participants | mouse-click saliency · OSIE | fixation prediction accuracy | Hurt |
| use BubbleView clicks instead of SALICON mouse movements | click data · OSIE | NSS vs fixations: clicks of 10 participants explain about 90% of fixations | Improved |
| apply image transformations (20 types) at test time | deep saliency models · transformed images | model performance | Hurt |
| augment saliency maps with annotated face/text/person regions | deep saliency models · MIT300 | NSS | Improved |
Closest papers by meaning
- Understanding and Visualizing Deep Visual Saliency Models (CVPR 2019 · similarity 0.94)
- A Deeper Look at Saliency: Feature Contrast, Semantics, and Beyond (CVPR 2016 · similarity 0.93)
- Revisiting Video Saliency Prediction in the Deep Learning Era (TPAMI 2019 · similarity 0.93)
- What Do Deep Saliency Models Learn about Visual Attention? (NeurIPS 2023 · similarity 0.93)
- Revisiting Video Saliency: A Large-Scale Benchmark and a New Model (CVPR 2018 · similarity 0.92)
- CaRDiff: Video Salient Object Ranking Chain of Thought Reasoning for Saliency Prediction with Diffusion (AAAI 2025 · similarity 0.90)
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.
