YYYong Yan

Yong Yan

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Yong Yan has a total of 26 co-authors. Here are the top co-authors by number of shared publications:

  • Xuejun Liu

    College of Information Engineering Beijing Institute of Petrochemical Technology Beijing China

    4 shared publications
  • Jiacheng Guo

    Beijing Institute of Petrochemical Technology, Department of Information Engineering, Beijing, China

    3 shared publications
  • Yun Sha

    Beijing Institute of Petrochemical Technology, Department of Information Engineering, Beijing, China

    3 shared publications
  • Jian-Ping Chen

    College of Information Engineering, Beijing Institute of Petrochemical Technology, 19 Qingyuan North Road, Daxing District, Beijing, China

    2 shared publications
  • Shuo Zhang

    Beijing Institute of Petrochemical Technology, Department of Information Engineering, Beijing, China

    2 shared publications
  • Yinan Jiang

    Beijing Institute of Petrochemical Technology, Department of Information Engineering, Beijing, China

    2 shared publications

Latest Publications

  • Traffic sign recognition using optimized YOLO network based on multi-scale feature extraction and attention mechanism

    by Yong Yan, Haoran Zhao, Yunpeng Li, Xuejun Liu, Weiwei Wang, Jiacheng Guo, Shuo Zhang, Yun Sha, Xuerui Wang, Yinan Jiang

    In the natural driving state, the autonomous vehicle needs to make its driving decision, which requires precise traffic sign recognition with different distances and angles according to road traffic rules. However, the change in the observation points of autonomous cameras causes traffic signs to exhibit visual characteristics of variable size and complex backgrounds. The coupling of multiple characteristics poses severe challenges for traffic sign identification and positioning. In this paper, a multi-scale feature extraction and coordinate network based on the YOLO framework (MFEC-YOLOv7) is proposed to achieve accurate traffic sign recognition in complex situations. In more detail, the feature for traffic signs at different scales is extracted and enhanced by fusing the pooling layer and group convolution in the convolutional architecture of the backbone network, thus improving the multi-scale feature perception capability. Furthermore, the coordinate attention (CA) module is introduced to pay attention to traffic signs under multiple interference factors. And the collaborative working mechanism of the bidirectional feature pyramid is brought in the neck network, and combined with the CA module to accurately locate the position information of features, to enhance the feature extraction ability of the model. Besides, depth separable convolution is adopted to significantly reduce the number of convolution parameters and improve the efficiency without reducing the effectiveness of feature extraction. The experimental results on the public data set TT100K and the self-built data set show that compared with the YOLOv7 method, MFEC-YOLOv7 improves the detection accuracy by 19.9%, reduces the amount of parameter calculation by 9.5%, and improves the speed by about 16.9%.

  • Two-stage graph matching point cloud registration method based on graph attention network

    by Jiacheng Guo, Xuejun Liu, Shuo Zhang, Yong Yan, Yun Sha, Yinan Jiang

    Accurate registration of point clouds acquired from different viewpoints can effectively reconstruct three-dimensional scenes, which plays a crucial role in industrial applications. Many learning-based methods rely on the extracted point cloud features to find the correspondence between two point clouds and thus achieve point cloud registration. However, repetitive geometric structures can make the local features of the learned point clouds highly similar, leading to the generation of incorrect correspondences. We propose the Two-Stage Graph Matching Point Cloud Registration Network (TSGM-Net). First, we design a dynamic graph-to-point (DGTP) module to learn the feature representation of the local graph of the point cloud to improve the recognition of local features. Then, edges are dynamically established by the Transformer and the introduced edge threshold λ, and the graph attention network is used to extract global features of the point cloud to consider the relationship among similar features in the topological structure. At the same time, scores are calculated from the three dimensions of the node itself, local and global, and summed for keypoint detection. Finally, we propose a two-stage graph matching method, where keypoints with highly similar features are divided into different point groups, and the correspondences of the point groups are established in the first-stage graph matching. The correspondence of points in corresponding point groups is established in the second-stage graph matching, which in turn reduces the impact of similar features on the accuracy of point cloud registration. Based on the benchmark dataset, we evaluate the performance of TSGM-Net in four different scenarios. Experimental results show that TSGM-Net can better handle point clouds with repetitive geometric structures and is competitive with state-of-the-art methods.


  • Video spatio-temporal generative adversarial network for local action generation

    by Xuejun Liu, Jiacheng Guo, Zhongji Cui, Ling Liu, Yong Yan, Yun Sha

    Generating action videos in future scenes based on static images can make computer vision systems to be better applied for video understanding and intelligent decision-making. However, current models pay more attention to the motion trend of the generated objects, and the processing effect on local details is not ideal. The local features of the generated video will have the problem of blurred frames and incoherent motion. This paper proposes a two-stage model, video spatio-temporal generative adversarial network (VSTGAN), which consists of two GAN networks, such as temporal network and spatial network (S-net). The model fully combines the advantages of CNNs, recurrent neural networks (RNNs), and GANs to decompose the complex spatiotemporal generation problem into temporal and spatial dimensions. Therefore, VSTGAN can focus on local features from the above dimensions respectively. In the temporal dimension, we propose an RNN unit, the convolutional attention unit (ConvAU), which uses the convolutional attention module to dynamically generate weights to update the hidden state. Thus, T-net uses the ConvAU to generate local dynamics. In the spatial dimension, S-net uses CNNs and attention modules to perform resolution reconstruction of the generated local dynamics for video generation. We build two small-sample datasets and validate our approach on these two new datasets and the KTH public dataset. The results show that our approach can effectively generate local details in future action videos and that the model performance on small-sample datasets is competitive with the state-of-the-art in video generation.


  • Retracted
    Open Access

    Feature Sequencing Method of Industrial Control Data Set Based on Multidimensional Evaluation Parameters

    by Xue-Jun Liu, Xiang-Min Kong, Xiao-Ni Zhang, Hai-Ying Luan, Yong Yan, Yun Sha, Kai-Li Li, Xue-Ying Cao, Jian-Ping Chen

    The industrial control data set has many features and large redundancy, which has a certain impact on the training speed and classification results of the neural network anomaly detection algorithm. However, features are independent of each other, and dimension reduction often increases the false positive rate and false negative rate. The feature sequencing algorithm can reduce this effect. In order to select the appropriate feature sequencing algorithm for different data sets, this paper proposes an adaptive feature sequencing method based on data set evaluation index parameters. Firstly, the evaluation index system is constructed by the basic information of the data set, the mathematical characteristics of the data set, and the association degree of the data set. Then, the selection model is obtained by the decision tree training with the data label and the evaluation index, and the suitable feature sequencing algorithm is selected. Experiments were conducted on 11 data sets, including Batadal data set, CICIDS 2017, and Mississippi data set. The sequenced data sets are classified by ResNet. The accuracy of the sequenced data sets increases by 2.568% on average in 30 generations, and the average time reduction per epoch is 24.143%. Experiments show that this method can effectively select the feature sequencing algorithm with the best comprehensive performance.


  • Open Access

    SingleNet: A Lightweight Convolutional Neural Network for Safety Detection of an Industrial Control System

    by Yun Sha, Jianping Chen, Jianwang Gan, Yong Yan, Xuejun Liu, Hao Wang

    The traditional industrial control security research mainly focuses on network intrusion detection or trapping system and lacks abnormal detection after intrusion, and the abnormal detection algorithm ability of the underlying operation data of industrial control is insufficient. The modern industrial control system is a side-cloud collaborative architecture that accesses the Internet, the edge side is usually an industrial computer with weak computing power, and the deep learning algorithm requires a lot of computing resources and is difficult to use directly on the edge side. In this paper, a lightweight convolutional neural network anomaly detection algorithm SingleNet suitable for the edge side of the industrial control system is proposed, which convolutes the data of each sensor for a period of time and calculates the feature correlation between points in the association calculation layer. Experimental results show that the accuracy rate on the oil depot dataset is increased from 73 to 99.4, the training time is shortened from 2 hours to 3 minutes, and the model size is compressed from 101MB to 1.6MB. The accuracy rate is improved from 87 to 99.2 on the Mississippi dataset, the training time is shortened from 15 minutes to 3 minutes, and the model size is compressed from 10.6MB to 1.63MB. The accuracy rate is improved from 85 to 99.4 on the Batadal dataset, the training time is shortened from 18 minutes to 3 minutes, and the model size is compressed from 15.5MB to 1.62MB. Compared with several lightweight algorithms recently proposed, SqueezeNet, MobileNet, and ShuffleNet, the proposed algorithm has significantly improved the performance indicators of training speed, accuracy, model size, and iteration time on the industrial control datasets. Both the training and testing of the algorithm can be done on the CPU, making it possible to apply deep learning to the edge side of the industrial control system.

  • Visual Clustering Analysis of Electricity Data Based on t-SNE

    by Feixiang Gong, Fanpeng Bu, Yufan Zhang, Yong Yan, Ruoyun Hu, Mingyu Dong

    The electricity consumption data recorded by smart meters contain plenty of knowledge that contribute to consumers categorization. However, traditional unsupervised methods for electricity data clustering suffer from the following problems: 1) They are not specialized for the high-dimensional electricity data, which have high computational complexity and poor clustering effects. 2) Most of the methods need to specify the model parameters in advance, and the general enumeration methods are inefficient. To solve these problems, we propose a novel Visual Clustering Analysis framework of Electricity Data (VCAED). To improve the validity of electricity data clustering and reduce clustering complexity, the t-SNE dimensionality reduction technology is used to reduce dimensions with information loss as less as possible, and visualization also makes the parameter selection process more intuitive. Relevant experiments on the real-world electricity consumption datasets validate the effectiveness and practicability of the proposed method.