Research map / Detection and segmentation / CNN features
Detect Globally, Refine Locally: A Novel Approach to Saliency Detection
Read the paperCheck 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 |
|---|---|---|---|
| Add the Contextual Weighting Module to the base network (ablation) | ResNet-50 saliency detector · ECSSD, THUR15K, HKU-IS, DUTS, DUT-OMRON | F-measure: ECSSD 0.861 -> 0.867; THUR15K 0.659 -> 0.667; HKU-IS 0.838 -> 0.840; DUTS 0.696 -> 0.716; DUT-OMRON 0.643 -> 0.661 | Helped |
| Add the Recurrent Module to the network with contextual weighting (ablation) | RLN saliency detector · ECSSD, THUR15K, HKU-IS, DUTS, DUT-OMRON | F-measure: ECSSD 0.867 -> 0.893; THUR15K 0.667 -> 0.702; HKU-IS 0.840 -> 0.875; DUTS 0.716 -> 0.760; DUT-OMRON 0.661 -> 0.712 | Helped |
| Add Boundary Refinement Network to the recurrent localization network (ablation) | RLN saliency detector · ECSSD, THUR15K, HKU-IS, DUTS | F-measure: ECSSD 0.893 -> 0.903; THUR15K 0.702 -> 0.716; HKU-IS 0.875 -> 0.882; DUTS 0.760 -> 0.768 | Helped |
| Add Boundary Refinement Network after RM (ablation) | RLN saliency detector · DUT-OMRON | F-measure: 0.712 -> 0.709 | Hurt |
| Use more recurrent time steps (ablation) | RLN on DUTS · DUTS | F-measure | Helped |
| Share parameters between recurrent time steps (ablation) | RLN on DUTS · DUTS | F-measure | Mixed |
| Apply top supervision at each time step rather than only at t=1 (ablation) | RLN on DUTS · DUTS | F-measure | Helped |
Closest papers by meaning
- Progressive Attention Guided Recurrent Network for Salient Object Detection (CVPR 2018 · similarity 0.92)
- BASNet: Boundary-Aware Salient Object Detection (CVPR 2019 · similarity 0.91)
- Recurrently Aggregating Deep Features for Salient Object Detection (AAAI 2018 · similarity 0.90)
- Non-Local Deep Features for Salient Object Detection (CVPR 2017 · similarity 0.90)
- Salient Object Detection Driven by Fixation Prediction (CVPR 2018 · similarity 0.90)
- Samba: A Unified Mamba-based Framework for General Salient Object Detection (CVPR 2025 · similarity 0.88)
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.