Wentao Yu

Wentao Yu

Auto-generated profile, based on publicly available information

Wentao Yu

School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

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

Representative Publications

View All Publications
4 October 2021 in IEEE Transactions on Cybernetics
Verified
Journal Article
Closed access
JBJing Bai
Jing Bai
WYWentao Yu
Wentao Yu
Zhu XiaoZhu Xiao
Zhu Xiao
VHVincent Havyarimana
Vincent Havyarimana
ARAmelia C. Regan
Amelia C. Regan
LJLicheng Jiao
Licheng Jiao
Convolutional neural networks (CNNs) have achieved remarkable performance in driver drowsiness detection based on the extraction of deep features of drivers’ faces. However, the performance of driver drowsiness detection methods decreases sharply when complications, such as illumination changes in the cab, occlusions and shadows on the driver’s face, and variations in the driver’s head pose, occur. In addition, current driver drowsiness detection methods are not capable of distinguishing between driver states, such as talking versus yawning or blinking versus closing eyes. Therefore, technical challenges remain in driver drowsiness detection. In this article, we propose a novel and robust two-stream spatial–temporal graph convolutional network (2s-STGCN) for driver drowsiness detection to solve the above-mentioned challenges. To take advantage of the spatial and temporal features of the input data, we use a facial landmark detection method to extract the driver’s facial landmarks from real-time videos and then obtain the driver drowsiness detection result by 2s-STGCN. Unlike existing methods, our proposed method uses videos rather than consecutive video frames as processing units. This is the first effort to exploit these processing units in the field of driver drowsiness detection. Moreover, the two-stream framework not only models both the spatial and temporal features but also models both the first-order and second-order information simultaneously, thereby notably improving driver drowsiness detection. Extensive experiments have been performed on the yawn detection dataset (YawDD) and the National TsingHua University drowsy driver detection (NTHU-DDD) dataset. The experimental results validate the feasibility of the proposed method. This method achieves an average accuracy of 93.4% on the YawDD dataset and an average accuracy of 92.7% on the evaluation set of the NTHU-DDD dataset.

30 January 2023 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
WYWentao Yu
Wentao Yu
SWSheng Wan
Sheng Wan
GLGuangyu Li
Guangyu Li
JYJian Yang
Jian Yang
CGchen gong
chen gong
Recently, Graph Convolutional Network (GCN) has been widely used in Hyperspectral Image (HSI) classification due to its satisfactory performance. However, the number of labeled pixels is very limited in HSI, and thus the available supervision information is usually insufficient, which will inevitably degrade the representation ability of most existing GCN-based methods. To enhance the feature representation ability, in this paper, a GCN model with contrastive learning is proposed to explore the supervision signals contained in both spectral information and spatial relations, which is termed Contrastive Graph Convolutional Network (ConGCN), for HSI classification. First, in order to mine sufficient supervision signals from spectral information, a semi-supervised contrastive loss function is utilized to maximize the agreement between different views of the same node or the nodes from the same land cover category. Second, to extract the precious yet implicit spatial relations in HSI, a graph generative loss function is leveraged to explore supplementary supervision signals contained in the graph topology. In addition, an adaptive graph augmentation technique is designed to flexibly incorporate the spectral-spatial priors of HSI, which helps facilitate the subsequent contrastive representation learning. The extensive experimental results on six typical benchmark datasets firmly demonstrate the effectiveness of the proposed ConGCN in both qualitative and quantitative aspects.

21 October 2021 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
JBJing Bai
Jing Bai
JRJunjie Ren
Junjie Ren
YYYujia Yang
Yujia Yang
Zhu XiaoZhu Xiao
Zhu Xiao
WYWentao Yu
Wentao Yu
LJLicheng Jiao
Licheng Jiao
Recently, optical remote-sensing images have been steadily growing in size, as they contain massive data and complex backgrounds. This trend presents several problems for object detection, for example, increased computation time and memory consumption and more false positives due to the complex backgrounds of large-scale images. Inspired by deep neural networks combined with time-frequency analysis, we propose a time-frequency analysis-based object detection method for large-scale remote-sensing images with complex backgrounds. We utilize wavelet decomposition to carry out a time-frequency transform and then integrate it with deep learning in feature optimization. To effectively capture the time-frequency features, we propose a feature optimization method based on deep reinforcement learning to select the dominant time-frequency channels. Furthermore, we design a discrete wavelet multiscale attention mechanism (DW-MAM), enabling the detector to concentrate on the object area rather than the background. Extensive experiments show that the proposed method of learning from time-frequency channels not only solves the challenges of large-scale and complex backgrounds, but also improves the performance compared to the original state-of-the-art object detection methods. In addition, the proposed method can be used with almost all object detection neural networks, regardless of whether they are anchor-based or anchor-free detectors, horizontal or rotation detectors.

23 June 2020 in IEEE Access
Verified
Journal Article
Open access
WYWentao Yu
Wentao Yu
JBJing Bai
Jing Bai
LJLicheng Jiao
Licheng Jiao
The application of deep learning techniques in background subtraction for VHR optical remote sensing videos holds the potential to facilitate multiple intelligent remote sensing processing tasks. However, existing methods on background subtraction for VHR optical remote sensing videos are still facing technical challenges. First, conventional CNN and other networks are limited by performance constraints. Second, existing background subtraction methods are mostly trained by natural videos due to the lack of VHR optical remote sensing video datasets. Third, VHR optical remote sensing videos have large scene sizes. In our article, we design a novel deep learning network via fully utilizing GAN and domain adaptation, which has the ability to measure and minimize the discrepancy between feature distributions of natural videos and VHR optical remote sensing videos so that the background subtraction performance for VHR optical remote sensing videos is improved significantly. Numerous experiments are conducted on the CDnet 2014 dataset and VHR optical remote sensing video dataset. Tremendous experiments demonstrate that our proposed method achieves an average FM of 0.8533, which reveals excellent performance on background subtraction.

21 February 2026 in ACM Transactions on Intelligent Systems and Technology
Verified
Journal Article
Open access
WYWentao Yu
Wentao Yu
SCShuo Chen
Shuo Chen
CGchen gong
chen gong
BHBo Han
Bo Han
GNGang Niu
Gang Niu
MSMasashi Sugiyama
Masashi Sugiyama
Recently, Graph Transformer (GT) models have been widely used in the task of Molecular Property Prediction (MPP) due to their high reliability in characterizing the latent relationship among graph nodes (i.e., the atoms in a molecule). However, most existing GT-based methods usually explore the basic interactions between pairwise atoms, and thus they fail to consider the important interactions among critical motifs (e.g., functional groups consisted of several atoms) of molecules. As motifs in a molecule are significant patterns that are of great importance for determining molecular properties (e.g., toxicity and solubility), overlooking motif interactions inevitably hinders the effectiveness of MPP. To address this issue, we propose a novel Atom-Motif Contrastive Transformer (AMCT), which not only explores the atom-level interactions but also considers the motif-level interactions. Since the representations of atoms and motifs for a given molecule are actually two different views of the same instance, they are naturally aligned to generate the self-supervisory signals for model training. Meanwhile, the same motif can exist in different molecules, and hence we also employ the contrastive loss to maximize the representation agreement of identical motifs across different molecules. Finally, in order to clearly identify the motifs that are critical in deciding the properties of each molecule, we further construct a property-aware attention mechanism into our learning framework. Our proposed AMCT is extensively evaluated on 10 popular benchmark datasets, and both quantitative and qualitative results firmly demonstrate its effectiveness when compared with the state-of-the-art methods.

12 August 2025 in TechRxiv
Verified
Preprint
Open access
WYWentao Yu
Wentao Yu
CGChen Gong
Chen Gong
BHBo Han
Bo Han
LFLixin Fan
Lixin Fan
QYQiang Yang
Qiang Yang
Real-world graphs are usually distributed across multiple organizations. Due to security regulations and privacy protection, distributed graphs are not allowed to be collected together and trained centrally. Therefore, Graph Federated Learning (GFL) has been proposed to carry out distributed collaborative graph representation learning while protecting the privacy of graph data. However, existing methods suffer from heterogeneity (i.e., non-independent and identically distributed graph data across different clients), which usually brings about unstable training, slow convergence, and degraded performance. Inspired by the idea of ‘seeking common ground while preserving individual differences’ in Personalized Federated Learning (PFL), we propose a novel graph Federated learning method by Integrating Commonality and Individuality (FedICI), which maximizes the consistency of commonalities across clients while minimizing the correlation of client-specific individualities among clients. Specifically, a spectral graph neural network with homophily bases and heterophily bases is employed to extract low-frequency components and high-frequency components from graph data on each client, where low-frequency and highfrequency components represent stable and discriminative signals of a client, respectively. On one hand, with the goal of maximizing the consistency of commonalities across different clients, common graph patterns are extracted from the low-frequency components on server side. On the other hand, with the aim of minimizing the correlation of individualities across heterogeneous clients, clientspecific graph patterns carried by high-frequency components are encouraged to be orthogonalized. Theoretically, we prove that our proposed FedICI achieves stable and fast convergence, where the convergence error decreases in proportion to the inverse square root of the number of local training epochs. Empirically, we conduct extensive experiments on six homophilic and five heterophilic graph datasets under both non-overlapping and overlapping settings. Experimental results validate the superiority of our method over eleven state-of-the-art methods. Notably, our FedICI outperforms the second-best method by an average margin of 3.36% on all heterophilic datasets, while also achieving more than threefold improvement in efficiency.