Research map / Detection and segmentation / CNN features
SPSN: Superpixel Prototype Sampling Network for RGB-D Salient Object Detection
Minhyeok Lee, Chaewon Park, Suhwan Cho, Sangyoun Lee
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 |
|---|---|---|---|
| add depth encoder to RGB-only baseline (ablation) | RGB baseline · NJU2K (train: DUT-RGBD+NJU2K+NLPR) | S-measure: .836 -> .863 | Improved |
| add reliance selection module (RSM) to RGB+depth encoders (ablation) | RGB+depth baseline · NJU2K (train: DUT-RGBD+NJU2K+NLPR) | S-measure: .863 -> .888 | Improved |
| add FFM/PGM/PSNM prototype sampling to RGB stream (ablation) | RGB baseline · NJU2K (train: DUT-RGBD+NJU2K+NLPR) | S-measure: .836 -> .903 | Improved |
| add prototype sampling streams to both RGB and depth (ablation) | RGB+depth baseline · NJU2K (train: DUT-RGBD+NJU2K+NLPR) | S-measure: .863 -> .912 | Improved |
| add RSM on top of prototype sampling in both streams (ablation) | prototype-sampling model · NJU2K (train: DUT-RGBD+NJU2K+NLPR) | S-measure: .912 -> .918 | Improved |
| add RSM on top of prototype sampling in both streams (ablation) | prototype-sampling model · DES (train: DUT-RGBD+NJU2K+NLPR) | S-measure: .922 -> .937 | Improved |
| add RSM on top of prototype sampling in both streams (ablation) | prototype-sampling model · NJU2K (train: DUT-RGBD+NJU2K+NLPR) | MAE: .035 -> .032 | Improved |
| add RSM on top of prototype sampling in both streams (ablation) | prototype-sampling model · SIP (train: DUT-RGBD+NJU2K+NLPR) | MAE: .047 -> .042 | Improved |
| add prototype sampling to RGB-only baseline (ablation) | RGB baseline · DES (train: DUT-RGBD+NJU2K+NLPR) | S-measure: .775 -> .905 | Improved |
| vary number of superpixels N_S away from 100 (ablation) | SPSN · NJU2K, STERE, DES | MAE and F-measure | Mixed |
| add depth encoder to RGB-only baseline (ablation) | RGB baseline · SIP (train: DUT-RGBD+NJU2K+NLPR) | S-measure: .738 -> .796 | Improved |
Closest papers by meaning
- Specificity-Preserving RGB-D Saliency Detection (ICCV 2021 · similarity 0.89)
- A Single Stream Network for Robust and Real-time RGB-D Salient Object Detection (ECCV 2020 · similarity 0.89)
- Calibrated RGB-D Salient Object Detection (CVPR 2021 · similarity 0.88)
- RGB-D Salient Object Detection via 3D Convolutional Neural Networks (AAAI 2021 · similarity 0.88)
- Promoting Saliency From Depth: Deep Unsupervised RGB-D Saliency Detection (ICLR 2022 · similarity 0.88)
- Alignment-Free RGB-T Salient Object Detection: A Large-Scale Dataset and Progressive Correlation Network (AAAI 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.
