Skip to content
Open access

DGQ-YOLO: Depth-Guided Quality-Aware detection with Pseudo-Depth for aerial imagery

Aug 2026 · Journal of King Saud University: Computer and Information Sciences · Vol 38 · 1 citation · 49 references

TL;DR

DGQ-YOLO, an RGB-dominant detector that introduces pseudo-depth at the backbone and neck at the backbone and neck is proposed, which improves pseudo-depth-assisted aerial detection while retaining detector-level efficiency, although online throughput remains constrained by pseudo-depth generation.

Abstract

Small, densely distributed, and occluded objects make UAV-view detection difficult because RGB-only detectors represent scene geometry only implicitly. Monocular pseudo-depth provides geometric cues without an additional sensor, but its derivation from RGB introduces local errors and estimator-dependent variation. We propose DGQ-YOLO, an RGB-dominant detector that introduces pseudo-depth at the backbone and neck. The Depth-Quality-Aware Adapter (DQAdapter) applies depth-conditioned spatial and channel gates through a zero-initialized residual path. The Consistency-Aware Geometric Feature Pyramid Network (CAGFPN) uses explicit RGB–depth discrepancy features to refine multi-scale representations. Across three random seeds on pseudo-depth-augmented VisDrone2019-DET, DGQ-YOLO-n improves test-set mAP50\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\textrm{mAP}_{50}$$\end{document} from 25.54% to 27.89% and mAP50:95\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\textrm{mAP}_{50:95}$$\end{document} from 14.07% to 15.64%. Controlled perturbation and gate-intervention experiments support reliability-sensitive modulation of the depth contribution. After dataset-specific retraining on UAVDT, DGQ-YOLO-n gains 2.14 mAP50:95\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\textrm{mAP}_{50:95}$$\end{document} points over its RGB baseline. The added modules increase parameter counts by approximately 3% and GFLOPs by less than 10%. DGQ-YOLO therefore improves pseudo-depth-assisted aerial detection while retaining detector-level efficiency, although online throughput remains constrained by pseudo-depth generation.

Read PDF

Similar papers

Preprint Sep 2026

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

Method is proposed, a reliability-aware geometry distillation framework developed for RGB-D SOD benchmarks without using dataset-provided depth during training or inference, and achieves the best or tied-best result in 26 of 36 metric-dataset comparisons against ten recent RGB-D SOD methods.

Xue-Hao Wang, Jia-Xin Hua, Run-Mei Li et al. · 0 citations
Preprint Sep 2026

NIDAR: NIR-Guided Intrinsic Decomposition for Scalable Scene-Agnostic LiDAR Intensity Reconstruction

LiDAR return intensity provides complementary surface-response cues for robotic perception and state estimation, yet many simulation pipelines omit it or reproduce it using reconstruction methods that require real intensity supervision and per-scene optimization. These requirements increase data-collection and fitting...

Jun-Jie Zhang, Jie Yin, Ke Qian et al. · 0 citations
Conference Aug 2026

FGP-BEVFusion for Lightweight Camera–LiDAR 3D Object Detection

Reliable 3D object detection is essential for autonomous-driving perception in complex traffic environments. Camera–LiDAR fusion combines dense visual semantics with precise point-cloud geometry to improve detection robustness. BEVFusion is a competitive bird’s-eye-view (BEV) fusion framework, but its high-capacity ima...

Zhi-Bin Shen, Cheng-Xi Huang, Guang-Lei Huo et al. · 0 citations
Open access 2026

Joint-MVSNet: Geometry-Controlled 2-D/3-D Interaction for Aerial Multiview Stereo Reconstruction

Learning-based aerial multiview stereo (MVS) has advanced large-scale 3-D reconstruction from remote-sensing imagery; however, reliable depth inference remains difficult in urban scenes with large relief variation, limited view overlap, occlusion, weak texture, repeated patterns, and radiometric inconsistency. These fa...

Liang-Yun Du, Bai-Yi Zong, Bo-Han Chen et al. · 0 citations
Open access 2026

SL-YOLO: Scale- and Lighting-Aware Object Detection for Challenging UAV Imagery

Complex illumination, perspective-induced scale variation, and repeated downsampling degrade object representations in unmanned aerial vehicle (UAV) imagery and make small-object localization highly sensitive to pixel-level shifts. To address these issues, we propose Scale- and Lighting-Aware YOLO (SL-YOLO), a lightwei...

Heng-Sheng Cai, Yuan-Lin Wang, Li Zhao et al. · 0 citations
Open access Oct 2026

SDU-G2A: Self-Derived Uncertainty Guidance for Ground-to-Aerial Remote-Sensing Image Synthesis with Probabilistic BEV Footprints

Urban planning and land-use monitoring require not only semantic land-cover maps, but also image-like overhead observations that preserve the spatial organization of urban scenes. Conventional remote-sensing segmentation and image-generation methods usually operate by obtaining available overhead imagery or in reliable...

Jie Liu, Gong-Jian Wen, Wei-Wei Wen et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.