Research map / Learning theory and algorithms / Prediction and decision losses
Revisiting Saliency Metrics: Farthest-Neighbor Area Under Curve
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 farthest-neighbor negatives rather than S-AUC negatives | Saliency evaluation · Toronto | FN-AUC: 0.515 -> 0.607 | Helped |
| Use global Gaussian smoothing instead of random-noise jitter | AUC computation · Quantized example map | AUC: 0.774 -> 0.790 | Helped |
| Vary the training ground-truth sigma (ablation) | CNN saliency evaluation · Across datasets | average deviation: CC 0.018; NSS 0.116; AUC-J 0.002; AUC-B 0.011; S-AUC 0.007; FN-AUC 0.008 | Mixed |
| Set the number of farthest neighbors to all other images (ablation) | FN-AUC · Saliency evaluation | FN-AUC | No clear effect |
Closest papers by meaning
- Spatially Binned ROC: A Comprehensive Saliency Metric (CVPR 2016 · similarity 0.86)
- Sanity Checks for Saliency Metrics (AAAI 2020 · similarity 0.85)
- Advancing Saliency Ranking with Human Fixations: Dataset Models and Benchmarks (CVPR 2024 · similarity 0.84)
- Shallow and Deep Convolutional Networks for Saliency Prediction (CVPR 2016 · similarity 0.84)
- Deep Unsupervised Saliency Detection: A Multiple Noisy Labeling Perspective (CVPR 2018 · similarity 0.84)
- End-To-End Saliency Mapping via Probability Distribution Prediction (CVPR 2016 · similarity 0.84)
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.