Research map / Detection and segmentation / CNN features
A2dele: Adaptive and Attentive Depth Distiller for Efficient RGB-D Salient Object Detection
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 |
|---|---|---|---|
| Add adaptive depth distillation (ablation) | RGB stream · DUT-RGBD | Fβ: .873 -> .886 | Helped |
| Add adaptive depth distillation (ablation) | RGB stream · NJUD | Fβ: .848 -> .867 | Helped |
| Add adaptive depth distillation (ablation) | RGB stream · NLPR | Fβ: .850 -> .870 | Helped |
| Add adaptive depth distillation (ablation) | RGB stream · STEREO | Fβ: .860 -> .877 | Helped |
| Add attentive depth distillation to adaptive depth distillation (ablation) | RGB stream · DUT-RGBD | Fβ: .886 -> .892 | Helped |
| Add attentive depth distillation to adaptive depth distillation (ablation) | RGB stream · NJUD | Fβ: .867 -> .874 | Helped |
| Add attentive depth distillation to adaptive depth distillation (ablation) | RGB stream · NLPR | Fβ: .870 -> .878 | Helped |
| Add attentive depth distillation to adaptive depth distillation (ablation) | RGB stream · STEREO | Fβ: .877 -> .884 | Helped |
| Replace adaptive depth weighting with fixed λ=0.3 (ablation) | RGB stream with depth distillation · DUT-RGBD, NJUD, NLPR, and STEREO | Fβ: DUT-RGBD: .883 vs .886; NJUD: .862 vs .867; NLPR: .863 vs .870; STEREO: .869 vs .877 | Hurt |
| Replace adaptive depth weighting with fixed λ=0.5 (ablation) | RGB stream with depth distillation · DUT-RGBD, NJUD, NLPR, and STEREO | Fβ: DUT-RGBD: .884 vs .886; NJUD: .863 vs .867; NLPR: .869 vs .870; STEREO: .875 vs .877 | Hurt |
| Replace adaptive depth weighting with fixed λ=0.7 (ablation) | RGB stream with depth distillation · DUT-RGBD, NJUD, NLPR, and STEREO | Fβ: DUT-RGBD: .863 vs .886; NJUD: .844 vs .867; NLPR: .843 vs .870; STEREO: .852 vs .877 | Hurt |
| Train with depth as privileged information, then use only RGB at test time | A2dele RGB stream · Five RGB-D benchmarks | Fβ | Helped |
Closest papers by meaning
- RGB-D Salient Object Detection via 3D Convolutional Neural Networks (AAAI 2021 · similarity 0.92)
- Promoting Saliency From Depth: Deep Unsupervised RGB-D Saliency Detection (ICLR 2022 · similarity 0.91)
- DVSOD: RGB-D Video Salient Object Detection (NeurIPS 2023 · similarity 0.91)
- DiMSOD: A Diffusion-Based Framework for Multi-Modal Salient Object Detection (AAAI 2025 · similarity 0.89)
- Alignment-Free RGB-T Salient Object Detection: A Large-Scale Dataset and Progressive Correlation Network (AAAI 2025 · similarity 0.89)
- DFormer: Rethinking RGBD Representation Learning for Semantic Segmentation (ICLR 2024 · 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.