Faculty Profiles - ZHANG Lifeng

Papers - ZHANG Lifeng

Division display21 - 40 of about 223／ All the affair displays >>

A novel iron metal organic framework-based nanocomposite for the synergistic degradation and adsorption of organic pollutants and Cr (Ⅵ) under visible light Reviewed International journal

Song X., Dai R., Zhu T., Zhang S., Liu J., Zhang S., Zhang L.

Journal of Cleaner Production 456 2024.06

YOLOv8MS: Algorithm for Solving Difficulties in Multiple Object Tracking of Simulated Corn Combining Feature Fusion Network and Attention Mechanism Reviewed International journal

Gao Y., Li Z., Li B., Zhang L.

Agriculture (Switzerland) 14 ( 6 ) 2024.06

Single Object Tracking Based on Enhanced Cross Attention Reviewed International journal

Ziyang Qi, Lifeng Zhang

Proceedings of the 12th IIAE International Conference on Industrial Application Engineering 2024 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 25 - 33 2024.03

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan matsue 2024.03.26 - 2024.03.30

The proposal and development of Siamese Network have led to the advancement of single object tracking algorithms that take into account real-time performance while improving performance results. However, based on the features extracted by the neural network, the traditional Siamese Network method only performs the correlation operation on them, so the tracker accuracy and robustness are deficient when there are occlusions and scale changes in the tracking scene. This paper proposes to use the Attention Head Network to replace the similarity calculation operation in the Siamese Network to calculate the cross features and enhance the model discriminability; and based on this, the cross features are enhanced to compensate for the lack of cross feature fusion. Experimental validation on the OTB100 and VOT2016 datasets yields great performance. The experiment results of both datasets are improved compared to the baseline tracker show good performance in tracking scenarios such as occlusion.

DOI： 10.12792/iciae2024.006

Other Link： https://www2.ia-engineers.org/conference/index.php/iciae/iciae2024/paper/view/3026

Visual Object Tracking with Appearance Model Updating and Tracking Reinitialization Reviewed International journal

Yuya Sun, Miaomiao Zhu(Changshu Institute of Technology), Shengrong Gong, Lifeng Zhang

Proceedings of the 12th IIAE International Conference on Industrial Application Engineering 2024 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 83 - 88 2024.03

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan matsue 2024.03.26 - 2024.03.30

Visual object trackers based on siamese network have performed exceptionally on multiple datasets, but their robustness are still influenced by the appearance deformation and loss of the object. Firstly, to adapt to changes in appearance, we update appearance model by replacing initial object templates with current object appearance features. Referring to deep reinforcement learning, this process can be modeled to a Markov decision process. So we introduce proximal policy optimization algorithm (PPO) to optimize it. Secondly, to further deal with the inevitable tracking failure, a tracking reinitialization is proposed to speculate whether the object is lost or not. If the object is believed to be lost, the local siamese tracker would be suspended and a global tracker will be launched to search the target object. Finally, we compare the proposed tracker with some popular methods on dataset OTB-2013, OTB-2015 and GOT-10k, confirming its effectiveness and efficiency by the experimental results.

DOI： 10.12792/iciae2024.015

Other Link： https://www2.ia-engineers.org/conference/index.php/iciae/iciae2024/paper/view/3029

Boundary-Aware Multi-modal Mirror Segmentation Reviewed International journal

Haonan Tang, Shuhan Chen(Yangzhou University), Lifeng Zhang

Proceedings of the 12th IIAE International Conference on Industrial Application Engineering 2024 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 34 - 37 2024.03

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan matsue 2024.03.26 - 2024.03.30

Mirror segmentation is a challenging task in the field of computer vision, with the difficulty being the system's struggle to distinguish between mirror reflections and actual objects in the background of an image. Traditional methods typically rely on the contextual relationships in the image and the reflective properties of mirrors for identification. However, these approaches are susceptible to the variability of mirror placement and the randomness of objects in the scene, consequently limiting the system's recognition capabilities. Moreover, due to the high similarity between the boundaries and the background, these methods often face issues of unclear or erroneous segmentation boundaries. Research has found that mirrors exhibit more pronounced features in depth images. Based on this discovery, we propose a boundary-aware multi-modal mirror segmentation method. This method involves the use of a multi-modal feature fusion module to integrate cross-modal feature. A boundary guidance module is also employed, utilizing boundary and high-frequency information to optimize the segmentation results. Compared to traditional image segmentation methods, our multi-modal approach demonstrates significant advantages in handling mirror segmentation tasks in complex scenes, particularly in terms of boundary clarity and segmentation accuracy.

DOI： 10.12792/iciae2024.007

Other Link： https://www2.ia-engineers.org/conference/index.php/iciae/iciae2024/paper/view/3053

Optimizing Speech Emotion Recognition with Hilbert Curve and convolutional neural network Reviewed International journal

Yang Z., Zhou S., Zhang L., Serikawa S.

Cognitive Robotics 4 30 - 41 2024.01

Magnetic elastomer composites with tunable magnetization behaviors for flexible magnetic transducers Reviewed International journal

Wang Y., Wang Z., Liu X., Zhang L., Li Q., Abdallaha D., Song X., Zhang L., Sun H., Liu J.

Polymer Composites 2024.01

FSANet Stereo3D Binocular Detection Network Reviewed International journal

Yuliang Gao, Zhen Li, Aoran Xi, Yuting Wang, Lifeng Zhang

Proceedings of the 11th IIAE International Conference on Industrial Application Engineering 2023 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 110 - 113 2023.03

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan Okinawa 2023.03.26 - 2023.03.30

Traditional binocular 3D object detection uses ResNet50 to extract multi-level features, which has the problems of lack of adaptability and loss of information. Inspired by YOLOV5, this work uses the Focus module to perform down-sampling to replace the ResNet50 network to extract multi-level information. On the other hand, the cost-volume are reweighted to get better depth. ASPP models are used for multi-dimensional object detection.The above constitutes the FSANet network.In the Kitti dataset, better results were obtained.

DOI： 10.12792/iciae2023.022

Other Link： https://www2.ia-engineers.org/conference/index.php/iciae/iciae2023/paper/view/2772

Deep Learning based Single-view 3D Reconstruction: A Survey Reviewed International journal

Wenjing Bian, Dongren Liu, Yaoyao Chen, Lifeng Zhang

Proceedings of the 11th IIAE International Conference on Industrial Application Engineering 2023 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 196 - 202 2023.03

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan Okinawa 2023.03.26 - 2023.03.30

Reconstruction 3D objects from a single view is an important research direction in computer vision. With the development of deep learning technology, a single-view 3D reconstruction based on deep learning has made remarkable progress in recent years. In reviewing 3D object reconstruction based on deep learning, firstly, the research progress of 3D reconstruction methods based on deep learning is analyzed in detail according to different representations of 3D objects, i.e., point cloud, voxel, and mesh. Secondly, the standard 3D reconstruction datasets and evaluation criteria are summarized. Finally, the challenges of 3D object reconstruction based on deep learning are summarized.

DOI： 10.12792/iciae2023.035

Other Link： https://www2.ia-engineers.org/conference/index.php/iciae/iciae2023/paper/view/2783

Research on Vegetable Growth Condition based on Computer vision Reviewed International journal

Yaoyao Chen, Zijun Yang, Yuting Wang, Aoran Xi, Wenjing Bian, Lifeng Zhang

Proceedings of the 11th IIAE International Conference on Industrial Application Engineering 2023 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 137 - 142 2023.03

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan Okinawa 2023.03.26 - 2023.03.30

This study is based on computer vision technology for soilless cultivation of vegetables. Traditional soilless culture requires human to judge the growth status of vegetables, while this study uses STM32 single-chip microcontroller as a microcontroller for environmental data collection and computer vision technology for intelligent detection of the growth status of vegetables. Finally, a comprehensive analysis of the most suitable growing conditions for vegetables. This paper focuses on the leaves of vegetables. Brassica chinensis was used as the object of study. The vegetable leaf area values were derived by different image pre-processing, segmentation, and feature extraction. For Brassica diseases, this study constructs a vegetable disease identification model based on an improved MobileNet-V2 network. This model is compared with similar network models and the speed is faster.

DOI： 10.12792/iciae2023.026

Other Link： https://www2.ia-engineers.org/conference/index.php/iciae/iciae2023/paper/view/2772

Pump Fault Diagnosis Based on Time Frequency Image and Parameters Improved CNN-LSTM Network Reviewed International journal

Aoran Xi, Yueping Mo, Tian Wen, Yuting Wang, Zhen Li, Yaoyao Chen, Lifeng Zhang

Proceedings of the 11th IIAE International Conference on Industrial Application Engineering 2023 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 129 - 136 2023.03

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan Okinawa 2023.03.26 - 2023.03.30

This paper combines the fault feature extraction and intelligent diagnosis of rolling bearings, and proposes a fault diagnosis model that combines 2D images with convolutional neural network (CNN) and long short memory network (LSTM), and optimizes the parameters. The different fault signals of rolling bearings are processed, and the processing capabilities of short-time Fourier, Wigner Ville distribution and wavelet transform for bearing signals are studied and compared. The Softmax function is used to realize multi fault pattern recognition and complete fault diagnosis. The experimental results show that the time-frequency image formed by wavelet transform has better performance on the eigenvalues of vibration signals of rotating bearings than other time-frequency image processing methods.

DOI： 10.12792/iciae2023.025

Other Link： https://www2.ia-engineers.org/conference/index.php/iciae/iciae2023/paper/view/2776

Marine Life Detection Algorithm Based on Deep Convolutional Generative Adversarial Network and ResNeXt50 with FPN Faster R-CNN Network Reviewed International journal

Yuting Wang, Seiichi Serikawa, Wenwen Shen, Lifeng Zhang, Zijun Yang, Tian Wen, Aoran Xi, Yaoyao Chen

Proceedings of the 11th IIAE International Conference on Industrial Application Engineering 2023 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 121 - 128 2023.03

Language：English Publishing type：Research paper (international conference proceedings)

Japan Okinawa 2023.03.26 - 2023.03.30

The development and utilization of marine resources has always been a focus of attention, and a prerequisite for the rational development of marine resources is the understanding of the distribution of biological resources in the ocean, so the detection of marine life has become an inevitable challenge in the development of marine resources. The acquisition of marine life images by underwater cameras is particularly difficult due to the complexity of the underwater terrain and the difficulty of underwater imaging. Therefore, the detection of marine life often faces problems of difficulties in dataset acquisition, scarcity of data in the dataset and low quality of images. To address the above problems, a marine life detection algorithm based on the combination of deep convolutional generative adversarial network and ResNeXt50 with FPN Faster R-CNN network is proposed in this paper. The algorithm uses Faster R-CNN as the basic framework for marine life detection and uses deep convolutional generative adversarial network to generate data for small dataset to effectively expand the dataset without the difficulty of underwater image acquisition. For the feature extraction network, we use ResNeXt50, which employs the idea of group convolution, as the basic network for feature extraction. Then, the network is combined with feature pyramid network to enhance the feature extraction capability. The expanded dataset fed into the ResNeXt50 with FPN Faster R-CNN target detection network for training. It can be demonstrated that the detection results of expanded dataset generated by DCGAN in ResNeXt50 with FPN Faster R-CNN network is better compared to the original dataset and the Resnet50 with FPN Faster R-CNN network.

DOI： 10.12792/iciae2023.024

Other Link： https://www2.ia-engineers.org/conference/index.php/iciae/iciae2023/paper/view/2772

Unsupervised Learning Based Indoor Personnel Behavior Pattern Mining Reviewed International journal

Tian Wen, Gaofei Sun, Yuting Wang, Aoran Xi, Lifeng Zhang

Proceedings of the 11th IIAE International Conference on Industrial Application Engineering 2023 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 114 - 120 2023.03

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan Okinawa 2023.03.26 - 2023.03.30

Indoor personnel behavior is an essential part of indoor environmental monitoring. Standard indoor personnel behavior recognition methods have problems, such as invasion of privacy and high equipment costs. This paper proposes a novel, low-privacy indoor-person behavior pattern analysis method. Three types of indoor person behavior recognition methods are formed through time and frequency domain analysis. Moreover, since the model will fail in new rooms, we use unsupervised learning to perform pattern recognition on the analyzed feature data to enhance the generalization and robustness of the model. The implementation results show that our algorithm can recognize people's behavior in rooms and that the improved pattern recognition model achieves 97% results, which is significantly better than more widely used models.

DOI： 10.12792/iciae2023.023

Other Link： https://www2.ia-engineers.org/conference/index.php/iciae/iciae2023/paper/view/2772

Facilitating enhanced microwave absorption properties of barium hexaferrite/polyaniline composites based on tunable interfacial polarization by rare earth doping Reviewed International journal

Liu X., Ji R., Yang M., Chen W., Chen H., Song X., Liu J., Zhang M., Zhang L.

Journal of Alloys and Compounds 937 2023.03

Proposal for Shoplifting Detection Using Skeletal Data Assuming Surveillance Camera Footage

Fujisakia Yujin, Zhang Lifeng

Proceedings of IIAE Annual Conference ( The Institute of Industrial Applications Engineers ) 2023 ( 0 ) 33 - 34 2023.01

Quantitative analysis of s-PB/SBR blend dispersion morphology using computer image processing-assisted Raman spectroscopic techniques Reviewed International journal

Ge M., Wu J., Hong Q., Zhang L., Zhang M., Yu L.

Analytical Methods 14 ( 40 ) 3982 - 3988 2022.09

Research on Fault Detection of Pump Rotating Bearing based on Deep Learning Reviewed International journal

Aoran Xi, Yueping Mo, Zijun Yang, Yuliang Gao, Shi Zhou, Yuting Wang, Lifeng Zhang

Proceedings of the 9th IIAE International Conference on Intelligent Systems and Image Processings 2022 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 155 - 161 2022.09

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan Kobe 2022.09.06 - 2022.09.10

Large-power water pump units play an important role in factory manufacturing and social life today. The bearing failure of water pump units seriously affects the operation of water pump equipment. In order to solve the problem of pump bearing fault classification, this paper compares four diagnosis methods based on deep learning. Three kinds of input signals come from feature extraction, wavelet transform time-frequency map preprocessing, and the original signals are employed. Then establish the deep learning models of CNN and CNN-LSTM for bearing fault diagnosis. Through comparison and analysis, the accuracy of using the time-frequency wavelet transform image as input can reach 100%. Finally, by comparing the training accuracy graph, confusion matrix, and scatter plot visualization, it is proved that the time-frequency graph of wavelet transform obtains the best result.

DOI： 10.12792/icisip2022.026

Other Link： https://www2.ia-engineers.org/conference/index.php/icisip/icisip2022/paper/view/2640

Optimize YoloStereo3D Binocular Detection Network by Attention Mechanism Reviewed International journal

Yuliang Gao, Zhen Li, Aoran Xi, Yuting Wang, Lifeng Zhang

Proceedings of the 9th IIAE International Conference on Intelligent Systems and Image Processings 2022 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 117 - 120 2022.09

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan Kobe 2022.09.06 - 2022.09.10

As the rapid development of computer vision, 3D object detection has become an indispensable method of perception for autonomous driving. LiDAR-only frameworks suffer from the sparsity of point clouds over the distance. While camera-only works suffer from the inaccurate depth measurement. Nevertheless, this paper proposed a LiDAR and camera fusion 3D detection framework, exploited and fused the spatiality of point clouds and the dense texture information of images together. Besides, we also utilized an automatic LiDAR and camera calibration method to ensure the accurate measurement while applied in embedded computation platform. Our experimental evaluation on KITTI dataset, the proposed work achieved good results on 3D detection accuracy and the automatic calibration method also made our work more stable and robust for different scenes.

DOI： 10.12792/icisip2022.020

Other Link： https://www2.ia-engineers.org/conference/index.php/icisip/icisip2022/paper/view/2676

The Fusion Thoughts in 3D Object Detection for Autonomous Driving: A Survey Reviewed International journal

Zhen Li, Shi Zhou, Yuliang Gao, Yuting Wang, Yuren Du, Seiichi Serikawa, Lifeng Zhang

Proceedings of the 9th IIAE International Conference on Intelligent Systems and Image Processings 2022 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 108 - 116 2022.09

Authorship：Last author Language：English Publishing type：Research paper (international conference proceedings)

Japan Kobe 2022.09.06 - 2022.09.10

DOI： 10.12792/icisip2022.019

Other Link： https://www2.ia-engineers.org/conference/index.php/icisip/icisip2022/paper/view/2652

Research on Two-stage Target Detection Algorithm and its Feature Extraction Network based on Deep Learning Reviewed International journal

Yuting Wang, Seiichi Serikawa, Wenwen Shen, Lifeng Zhang, Chao Yuan, Zhen Li, Shi Zhou

Proceedings of the 9th IIAE International Conference on Intelligent Systems and Image Processings 2022 ( The Institute of Industrial Applications Engineers ) 1 ( 1 ) 139 - 145 2022.09

Language：English Publishing type：Research paper (international conference proceedings)

Japan Kobe 2022.09.06 - 2022.09.10

Deep learning two-stage target detection algorithms have been widely used in the field of target detection relying on the advantages of high accuracy. This paper introduces four common two-stage target detection algorithms in detail: R-CNN, Fast R-CNN, Faster R-CNN and Mask R-CNN, and expounds their respective algorithm steps, advantages and disadvantages. At the same time, the feature extraction network that is often used in the current two-stage target detection algorithm is analyzed and summarized from two directions. One direction is to pursue image classification networks with deeper and more complex network structures and higher detection accuracy for feature extraction, such as AlexNet, VGGNet, GoogleNet, ResNet and DenseNet. Another direction is to solve the problem that the network parameters are too large and the detection speed is too slow, a lightweight neural network is proposed, hoping to reduce the parameter calculation amount of the neural network and speed up the detection speed of the network, such as SqueezeNet, MobileNet, ShuffleNet and GhostNet. This paper introduces the network structure of the above nine networks and their respective advantages and disadvantages from the above two directions, and provides a theoretical and method reference for the selection of the feature extraction network of the two-stage target detection algorithm.

DOI： 10.12792/icisip2022.024

Other Link： https://www2.ia-engineers.org/conference/index.php/icisip/icisip2022/paper/view/2634

　 PREV - NEXT 　

<ZHANG Lifeng>

Papers - ZHANG Lifeng