Research map / Detection and segmentation / CNN features
Optimizing the F-Measure for Threshold-Free Salient Object Detection
Kai Zhao, Shanghua Gao, Wenguan Wang, Ming-Ming Cheng
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 |
|---|---|---|---|
| train with FLoss instead of cross-entropy | DSS · ECSSD (MSRA-B) | MeanF: .889 -> .903 | Improved |
| train with FLoss instead of cross-entropy | DSS · ECSSD (MSRA-B) | MAE (lower better): .060 -> .050 | Improved |
| train with FLoss instead of cross-entropy | DSS · DUT-OMRON (MSRA-B) | MaxF: .761 -> .777 | Improved |
| train with FLoss instead of cross-entropy | Amulet · ECSSD (MSRA-B) | MaxF: .881 -> .894 | Improved |
| train with FLoss instead of cross-entropy | Amulet · SOD (MSRA-B) | MeanF: .776 -> .800 | Improved |
| train with FLoss instead of cross-entropy | DHS · HKU-IS (MSRA-B) | MaxF: .835 -> .859 | Improved |
| train with FLoss instead of cross-entropy | DHS · DUT-OMRON (MSRA-B) | MAE (lower better): .078 -> .079 | No clear effect |
| use relaxed F-measure loss instead of its log version (Log-FLoss) (ablation) | DSS · ECSSD (MSRA-B) | MeanF: .891 -> .903 | Improved |
| use relaxed F-measure loss instead of its log version (Log-FLoss) (ablation) | DSS · DUT-OMRON (MSRA-B) | MAE (lower better): .062 -> .067 | Hurt |
| use relaxed F-measure loss instead of its log version (Log-FLoss) (ablation) | DSS · ECSSD (MSRA-B) | MAE (lower better): .057 -> .050 | Improved |
| use FLoss instead of balanced cross-entropy (ablation) | DSS · ECSSD (MSRA-B) | MeanF: .890 -> .903 | Improved |
| use balanced cross-entropy instead of plain cross-entropy (ablation) | DSS · ECSSD (MSRA-B) | MeanF: .889 -> .890 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Amulet — Amulet: Aggregating Multi-Level Convolutional Features for Salient Object Detection (ICCV 2017) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- F³Net: Fusion, Feedback and Focus for Salient Object Detection (AAAI 2020 · similarity 0.88)
- Deep Embedding Features for Salient Object Detection (AAAI 2019 · similarity 0.87)
- Non-Local Deep Features for Salient Object Detection (CVPR 2017 · similarity 0.87)
- Recurrently Aggregating Deep Features for Salient Object Detection (AAAI 2018 · similarity 0.87)
- What Is and What Is Not a Salient Object? Learning Salient Object Detector by Ensembling Linear Exemplar Regressors (CVPR 2017 · similarity 0.86)
- Revisiting Salient Object Detection: Simultaneous Detection, Ranking, and Subitizing of Multiple Salient Objects (CVPR 2018 · similarity 0.86)
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.
