Research map / Detection and segmentation / CNN features
CNN-Based RGB-D Salient Object Detection: Learn, Select, and Fuse
Hao Chen, Youfu Li, Yongjian Deng, Guosheng Lin
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 depth-trained student stream instead of RGB teacher on depth input | zero-shot depth saliency · NJUD | F-measure: 0.598 -> 0.753 | Improved |
| initialize depth network with cross-modal distillation instead of RGB weights | depth saliency network · NJUD | F-measure: 0.780 -> 0.796 | Improved |
| train depth network from scratch with ground-truth side supervision | depth saliency network · NLPR/NJUD/STEREO | F-measure | Hurt |
| fine-tune the distilled student with ground-truth masks | depth saliency network · NLPR | F-measure: 0.784 -> 0.792 | Improved |
| fine-tune the distilled student with ground-truth masks | depth saliency network · STEREO | F-measure: 0.739 -> 0.737 | No clear effect |
| initialize RGB-D network with distillation instead of ImageNet VGG | RGB-D saliency network · NLPR | F-measure: 0.842 -> 0.872 | Improved |
| initialize RGB-D network with distillation instead of stage-wise training | RGB-D saliency network · NLPR | F-measure: 0.861 -> 0.872 | Improved |
| add cross-modal residual connections to CA-Fuse (ablation) | CA-Fuse block · NLPR | F-measure: 0.842 -> 0.867 | Improved |
| add RGB and depth branch supervision L_R and L_D (ablation) | CA-Fuse block · NLPR | F-measure: 0.867 -> 0.872 | Improved |
| add cross-level feature transmission to adjacent shallower layer (ablation) | CA-Fuse block · NJUD | F-measure: 0.862 -> 0.871 | Improved |
Closest papers by meaning
- Progressively Complementarity-Aware Fusion Network for RGB-D Salient Object Detection (CVPR 2018 · similarity 0.95)
- RGB-D Salient Object Detection with Cross-Modality Modulation and Selection (ECCV 2020 · similarity 0.94)
- Learning Selective Mutual Attention and Contrast for RGB-D Saliency Detection (TPAMI 2021 · similarity 0.92)
- Hierarchical Dynamic Filtering Network for RGB-D Salient Object Detection (ECCV 2020 · similarity 0.92)
- Specificity-Preserving RGB-D Saliency Detection (ICCV 2021 · similarity 0.91)
- DiMSOD: A Diffusion-Based Framework for Multi-Modal Salient Object Detection (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.
