Zhihong Zeng

Multimodal fusion and pseudo-labeling for enhanced weakly supervised camouflaged object detection

Danyang Yang, Zao Liu, Zhihong Zeng, Bei Cheng

cn (code pays fourni par la source)

Multimodal Fusion and Pseudo-Labeling for Enhanced Weakly-Supervised Camouflaged Object Detection

Danyang Yang, Zao Liu, Zhihong Zeng, Bei Cheng

cn (code pays fourni par la source)

DACE-Net: Detail-Aware Calibrated Enhancement Network for Building Change Detection

Fenglei Chen, Xiaoheng Tan, Zhihong Zeng, Hailin Cao et autres

As one of the critical tasks in urban change interpretation, building change detection aims to acquire large-scale spatial distribution change information of buildings in high-resolution remote sensing images. The false detection of pseudochanges remains a significant challenge impacting detection accuracy. Remote sensing …

cn (code pays fourni par la source)

Experimental Demonstration of Multi-User Full-Duplex RO-ISAC System

Shuhang Chen, Zhihong Zeng

A multi-user full-duplex retroreflective optical ISAC (RO-ISAC) system using wavelength division duplexing (WDD) and interference cancellation is proposed. Experimental results demonstrate the feasibility of multi-user joint sensing and communication in the RO-ISAC system.

cn (code pays fourni par la source)

Interference Mitigation for Faster-Than-Nyquist mCAP in Bandlimited VLC Using Variable GIM Scheme

Yungui Nie, Siming Mo, Xiaodi You, Qinghai Lu et autres

A novel variable gapped index modulation (VGIM) scheme is proposed to mitigate the IBI in FTN-mCAP for bandlimited VLC systems. Experimental results show that FTN-8CAP-VGIM achieves a substantial SE improvement of 108% compared with 8CAP.

cn (code pays fourni par la source)

Spatial Index Modulation Multiple Access for Multiuser MIMO-OWC Systems

Xin Zhong, Zhihong Zeng, Dengke Wang, Chen Chen et autres

This paper presents a multi-user multiple-input multiple-output optical wireless communication (MU-MIMOOWC) system employing single/dual-mode spatial index modulation multiple access (SM/DM-SIMMA). The proposed SM/DMSIMMA schemes enable efficient two-dimensional bit allocation through distinct assignment of spatial and constellation bits to different users. Through comprehensive …

cn, gb (code pays fourni par la source)

MambaSOD: Dual Mamba-driven cross-modal fusion network for RGB-D Salient Object Detection

Yue Zhan, Zhihong Zeng, Haijun Liu, Xiaoheng Tan et autres

hk, cn (code pays fourni par la source)

Semantic and consistent neural radiance field reconstruction method based on intrinsic decomposition via classification

Zhihong Zeng, Zongji Wang, Yuanben Zhang, Weinan Cai et autres

目的基于神经辐射场（neural radiance field，NeRF）的3D场景重建与新视角生成工作正受到研究者的广泛重视，然而现有的神经辐射场方法通常对给定的场景进行高度专门化的表征，且将场景的几何与外观表征为“混合场”，这对场景的几何与外观编辑、场景泛化和3D资源的使用造成了不便。方法提出了一个学习对象本征属性的神经辐射场分类网络，通过图像增强的方式去除高光和阴影，并使用分类的方式实现颜色分解，即从现实场景中提取室内场景语义级目标的本征属性，在此基础上进行神经辐射场的重建。提出了前点优胜模块与颜色分类模块。前点优胜模块在体渲染阶段优化射线代表的本征属性，从而提升神经辐射场的语义一致性；颜色分类模块在辐射场重建阶段，通过全连接网络进行本征属性的分类优化，提高辐射场的语义及视角间一致性。两个主要模块共同作用，使重建的辐射场具备良好的针对外观的泛化能力，可支持场景重上色、重光照以及针对阴影与高光的编辑等任务。结果相比于现有的基于神经辐射场的学习进行本征分解的Intrinsic NeRF方法，在Replica数据集中的充分实验表明，在有限的GPU显存和运行时间下，重建的本征属性神经辐射场具备语义及视角间一致性。针对提升语义一致性的前点优胜模块，本文方法在基线模型Semantic NeRF的基础上提高了4.1%，在未加入该模块的基础上提高了 3.9%。针对提升本征分解语义及视角间一致性的颜色分类模块，本文方法在Intrinsic NeRF的本征分解工作基础上提升了10.2%，在未加入颜色分类层的基础上提升了1.7%。结论本文方法构建的本征属性神经辐射场具备语义及视角间一致性，可描述复杂场景几何关系且具备良好外观泛化性。在场景重上色、重光照、阴影与高光的编辑等任务中取得了视角间一致的逼真效果。

FASI-Net: Frequency-Domain Information Assisted Semantic Interaction Network for Bitemporal Remote Sensing Images Change Detection

Fenglei Chen, Haijun Liu, Zhihong Zeng, Xiaoheng Tan

As a crucial approach for comprehending land surface changes, remote sensing images (RSIs) change detection methods based on deep learning have been extensively studied in recent years. Among these methods, the Siamese network-based method has demonstrated remarkable performance. However, existing approaches are …

cn (code pays fourni par la source)

RGB-Guided Depth Feature Enhancement for RGB–Depth Salient Object Detection

Zhihong Zeng, Jiahao He, Yue Zhan, Haijun Liu et autres

RGB-D (depth) Salient Object Detection (SOD) seeks to identify and segment the most visually compelling objects within a given scene. Depth data, known for their strong discriminative capability in spatial localization, provide an advantage in achieving accurate RGB-D SOD. However, recent research …

cn, hk (code pays fourni par la source)

Retroreflective Optical ISAC Supporting 3D Positioning in Indoor Environments

Haochuan Wang, Zhihong Zeng, Chen Chen, Sihua Shao et autres

A retroreflective optical ISAC (RO-ISAC) system supporting 3D positioning is proposed. This system used the transmitted signal and the reflected signal to measure the distance, which can further locate position of the target.

cn, us (code pays fourni par la source)

MambaSOD: Dual Mamba-Driven Cross-Modal Fusion Network for RGB-D Salient Object Detection

Yue Zhan, Zhihong Zeng, Haijun Liu, Xiaoheng Tan et autres

The purpose of RGB-D Salient Object Detection (SOD) is to pinpoint the most visually conspicuous areas within images accurately. While conventional deep models heavily rely on CNN extractors and overlook the long-range contextual dependencies, subsequent transformer-based models have addressed the issue to …