Wei Hong

Dr. Wei Hong

orcid

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 January 2025 in Sensors
Verified
Journal Article
Open access
PDF file
YGYuhan Gao
Yuhan Gao
PWPeng Wang
Peng Wang
XLXiaoyan Li
Xiaoyan Li
MSMengyu Sun
Mengyu Sun
RDRuohai Di
Ruohai Di
WHWei Hong
Wei Hong
Monocular 3D object detection refers to detecting 3D objects using a single camera. This approach offers low sensor costs, high resolution, and rich texture information, making it widely adopted. However, monocular sensors face challenges from environmental factors like occlusion and truncation, leading to reduced detection accuracy. Additionally, the lack of depth information poses significant challenges for predicting 3D positions. To address these issues, this paper presents a monocular 3D object detection method based on improvements to MonoCD, designed to enhance detection accuracy and robustness in complex environments. In order to effectively obtain and integrate depth information, this paper designs a multi-branch depth prediction with weight sharing module. Furthermore, an adaptive focus mechanism is proposed to emphasize target regions while minimizing interference from irrelevant areas. The experimental results demonstrate that MonoDFNet achieves significant improvements over existing methods, with AP3D gains of +4.09% (Easy), +2.78% (Moderate), and +1.63% (Hard), confirming its effectiveness in 3D object detection.

25 December 2025 in IEEE Transactions on Vehicular Technology
Verified
Journal Article
Closed access
WHWei Hong
Wei Hong
WHWei Hong
Wei Hong
RDRuohai Di
Ruohai Di
SJShoucai Jing
Shoucai Jing
CHChuchao He
Chuchao He
XZXiangmo Zhao
Xiangmo Zhao
The credibility of the perception system is a linchpin for the safety of Automated Vehicles (AVs). However, existing scenario-based testing approaches often overlook the reliability assessment of the perception system. To address this shortcoming, we propose BAGA, a Bayesian-based adaptive search method for high-value scenarios, which accelerates scenario-based simulations while identifying critical scenarios for the perception system. Concurrently, we design a sensitivity analysis method, BNSobol, based on the naive Bayes network. This method decouples the associations among selected trigger source sets and develops a mutation strategy that aligns with the physical characteristics of the perception system. Furthermore, we construct an objective function that integrates the tolerance boundaries of the credibility indicators. This function calculates the perception system's credibility within scenarios and identifies high-value scenarios. Ultimately, we develop integrated software for evaluating the credibility of the perception system, providing a universal solution for such assessments. Experimental results demonstrate that, compared to mainstream scenario generation methods and exhaustive search methods, our proposed method increases efficiency by 77.7% and 143.2%, respectively. It can effectively generate the high-value scenarios required for perception system testing, thereby facilitating the discovery and localization of credibility issues.