Skip to content
Conference

Collaborative Dehazing and Detection for Traffic Signs in Complex Foggy Scenes

Sep 2026 · 2026 9th International Conference on Computer Network, Electronic and Automation (ICCNEA) · pp. 1-6 · 0 citations · 10 references

Abstract

Traffic signs are particularly vulnerable to fog-induced contrast attenuation, edge blurring, and the loss of discriminative cues for small objects. This paper proposes a collaborative dehazing-and-detection framework based on DehazeFormer-S and YOLO11n. A differentiable Joint-Basic model is first constructed so that the detection loss can backpropagate through the detector and constrain the dehazing branch. Three complementary mechanisms are then introduced: traffic-sign region-of-interest (ROI)-constrained local reconstruction and edge preservation, fog-intensity-adaptive feature-wise linear modulation (FiLM), and a Fog-Dehaze Difference-guided Detail Reorganization module for the P3 branch (FDR-P3). FDR-P3 derives a spatial gate from the difference between foggy and dehazed images and selectively injects shallow details into P3 features to strengthen small-object representation. Experiments on a synthetic foggy traffic-sign dataset derived from TT100K_2021 show that the complete model achieves 73.92% precision, 65.82% recall, 72.21% mAP50, and 54.32% mAP50-95. The mAP50-95 improves by 1.71 percentage points over Joint-Basic. Ablation results demonstrate that the three modules provide complementary benefits for traffic-sign detection under foggy conditions.

View source

Similar papers

Open access Sep 2026

DEO-NET: A Small Object Detection Method for Foggy Low-Visibility Scenarios

DEO-NET is proposed, a fog-aware object detection framework tailored for transmission line insulator defect detection that attains superior overall performance and enhanced robustness across diverse foggy scenarios, while maintaining acceptable model complexity in terms of both parameter count and computational cost.

Yu-Qian Wang, Jun-Yan Wang, Bao-Xi Yuan et al. · 0 citations
Open access Sep 2026

RiDW-YOLO: A Low-Light Traffic Sign Detection Algorithm Integrating Illumination Enhancement

RiDW-YOLO, an improved detection algorithm based on YOLOv11n, provides a feasible solution for traffic-sign detection under low-light conditions and emphasizes the importance of balanced module design and frank discussion of current limitations.

Yin-Yin Li, Lei Liu, Fang-Zheng Tong et al. · 0 citations
Conference Sep 2026

A method for compressing drone-based small object detection models in complex outdoor scenarios

Detecting small targets such as forest fire points in complex outdoor scenes faces challenges including small target size and strong background interference, and is easily affected by motion blur, illumination changes, and complicated backgrounds. Small targets in remote sensing images typically have low pixel coverage...

Grace W.W. Gao · 0 citations
Open access Sep 2026

Hybrid object detection in low visibility underwater imagery

This study presents a multilevel analysis of object detection in underwater and aerial imagery captured under limited-visibility conditions. The physical causes of input-signal degradation, including wavelength-dependent absorption, scattering, non-uniform illumination, and low-contrast blur, are systematized, feature...

D. Moroz · 0 citations
Open access Sep 2026

DPFS-YOLO: Missed-Detection Alleviation and False-Detection Risk Suppression for Small Objects in Complex UAV Aerial Scenes

Experiments on the VisDrone2019-DET, HIT-UAV, and TinyPerson datasets demonstrate that the proposed DPFS-YOLO method effectively alleviates missed detections of small objects and reduces false-detection risk under complex background conditions.

Zong-Ran Yang, Run-Jie Liu, Fei Wang et al. · 1 citation

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