# When Depth Hurts: Reliability-Aware Geometry Distillation for Depth-Free RGB-D Salient Object Detection

arXiv: https://arxiv.org/abs/2609.03378 · HTML: https://mlsanity.com/paper/2609.03378/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 25 of its 41 references that are in the map.
- First posted on arXiv 3 Sep 2026.

## Builds on (most cited first, 25 of 25)

- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [FitNets: Hints for Thin Deep Nets](https://mlsanity.com/paper/1412.6550/) (2014)
- [Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data](https://mlsanity.com/paper/2401.10891/) (2024)
- [Paying More Attention to Attention: Improving the Performance of Convolutional Neural Networks via Attention Transfer](https://mlsanity.com/paper/1612.03928/) (2016)
- [Structure-measure: A New Way to Evaluate Foreground Maps](https://mlsanity.com/paper/1708.00786/) (2017)
- [Enhanced-alignment Measure for Binary Foreground Map Evaluation](https://mlsanity.com/paper/1805.10421/) (2018)
- [Coordinate Attention for Efficient Mobile Network Design](https://mlsanity.com/paper/2103.02907/) (2021)
- [EGNet:Edge Guidance Network for Salient Object Detection](https://mlsanity.com/paper/1908.08297/) (2019)
- [The Secrets of Salient Object Segmentation](https://mlsanity.com/paper/1406.2807/) (2014)
- [F3Net: Fusion, Feedback and Focus for Salient Object Detection](https://mlsanity.com/paper/1911.11445/) (2019)
- [Visual Saliency Transformer](https://mlsanity.com/paper/2104.12099/) (2021)
- [Salient Object Detection in the Deep Learning Era: An In-Depth Survey](https://mlsanity.com/paper/1904.09146/) (2019)
- [Rethinking RGB-D Salient Object Detection: Models, Data Sets, and Large-Scale Benchmarks](https://mlsanity.com/paper/1907.06781/) (2019)
- [Salient Object Detection via Integrity Learning](https://mlsanity.com/paper/2101.07663/) (2021)
- [UC-Net: Uncertainty Inspired RGB-D Saliency Detection via Conditional Variational Autoencoders](https://mlsanity.com/paper/2004.05763/) (2020)
- [JL-DCF: Joint Learning and Densely-Cooperative Fusion Framework for RGB-D Salient Object Detection](https://mlsanity.com/paper/2004.08515/) (2020)
- [RGB-D Salient Object Detection: A Survey](https://mlsanity.com/paper/2008.00230/) (2020)
- [VSCode: General Visual Salient and Camouflaged Object Detection with 2D Prompt Learning](https://mlsanity.com/paper/2311.15011/) (2023)
- [RGB-D Saliency Detection via Cascaded Mutual Information Minimization](https://mlsanity.com/paper/2109.07246/) (2021)
- [Learning Selective Mutual Attention and Contrast for RGB-D Saliency Detection](https://mlsanity.com/paper/2010.05537/) (2020)
- [RGB-D Salient Object Detection via 3D Convolutional Neural Networks](https://mlsanity.com/paper/2101.10241/) (2021)
- [HiDAnet: RGB-D Salient Object Detection via Hierarchical Depth Awareness](https://mlsanity.com/paper/2301.07405/) (2023)
- [VST++: Efficient and Stronger Visual Saliency Transformer](https://mlsanity.com/paper/2310.11725/) (2023)
- [Visual Saliency Detection Based on Multiscale Deep CNN Features](https://mlsanity.com/paper/1609.02077/) (2016)

## Abstract (arXiv)

Depth can resolve appearance ambiguity in RGB-D salient object detection (SOD), yet sensor depth is not uniformly reliable. Missing regions, blurred boundaries, and structural artifacts can propagate through multimodal fusion and make an RGB-D detector less accurate than its RGB-only counterpart. Existing quality-aware approaches regulate observed depth but remain dependent on the same potentially defective modality. We propose method, a reliability-aware geometry distillation framework developed for RGB-D SOD benchmarks without using dataset-provided depth during training or inference. A frozen Depth Anything V2 model serves only as a training-time teacher, transferring dense relative geometry, hierarchical spatial attention, and boundary structure to a compact edge-aware geometry branch. Pooled bidirectional interaction aligns geometry with appearance, and a pixel-wise reliability estimator selectively injects geometry that is compatible with the current RGB representation. The teacher is removed after training, leaving an RGB-only inference network. Trained on 2,985 RGB-mask pairs, method achieves the best or tied-best result in 26 of 36 metric-dataset comparisons against ten recent RGB-D SOD methods, including a 13.4% relative MAE reduction on ReDWeb-S. When retrained on DUTS-TR, it also improves the strongest prior F-measure by 4.2% on PASCAL-S, showing that the distilled geometry transfers beyond a particular sensor or dataset domain. Code will be released upon publication.