{"slug": "when-depth-hurts-reliability-aware-geometry-distillation-for-depth-free-rgb-d", "title": "When Depth Hurts: Reliability-Aware Geometry Distillation for Depth-Free RGB-D Salient Object Detection", "summary": "A new reliability-aware geometry distillation framework enables RGB-D salient object detection without using dataset-provided depth during training or inference, achieving the best or tied-best result in 26 of 36 metric-dataset comparisons against ten recent methods, including a 13.4% relative MAE reduction on ReDWeb-S. The method, developed by researchers and posted on arXiv (2609.03378v1), uses a frozen Depth Anything V2 model as a training-time teacher and removes it after training, leaving an RGB-only inference network.", "body_md": "arXiv:2609.03378v1 Announce Type: new\nAbstract: 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.", "url": "https://wpnews.pro/news/when-depth-hurts-reliability-aware-geometry-distillation-for-depth-free-rgb-d", "canonical_source": "https://arxiv.org/abs/2609.03378", "published_at": "2026-09-04 04:00:00+00:00", "updated_at": "2026-09-04 04:24:53.177109+00:00", "lang": "en", "topics": ["computer-vision", "machine-learning", "artificial-intelligence"], "entities": ["arXiv", "Depth Anything V2", "ReDWeb-S", "DUTS-TR", "PASCAL-S"], "alternates": {"html": "https://wpnews.pro/news/when-depth-hurts-reliability-aware-geometry-distillation-for-depth-free-rgb-d", "markdown": "https://wpnews.pro/news/when-depth-hurts-reliability-aware-geometry-distillation-for-depth-free-rgb-d.md", "text": "https://wpnews.pro/news/when-depth-hurts-reliability-aware-geometry-distillation-for-depth-free-rgb-d.txt", "jsonld": "https://wpnews.pro/news/when-depth-hurts-reliability-aware-geometry-distillation-for-depth-free-rgb-d.jsonld"}}