Dongyu He
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Dongyu He has a total of 19 co-authors. Here are the top co-authors by number of shared publications:
- Nianyin Zeng
Xiamen University
6 shared publications - Peishu Wu
Department of Instrumental and Electrical Engineering, Xiamen University, Xiamen, China
6 shared publications - Baoping Xiong
School of Computer Science and Mathematics, Fujian University of Technology, Fuzhou, China
5 shared publications - Feng Guo
Fujian Key Laboratory of Automotive Electronics and Electric Drive, Fujian University of Technology, Fuzhou 350118, China
5 shared publications - Zhenhua Gan
School of Computer Science and Mathematics, Fujian University of Technology, Fuzhou, China
5 shared publications - Fumin Zou
Fujian Key Laboratory of Automotive Electronics and Electric Drive, Fujian University of Technology, Fuzhou 350118, China
3 shared publications
Latest Publications
- 30 December 2025
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 37 (2) , 026107
A novel microarray detection method based on deep learning with edge computing
Traditional microarray scanners typically rely on high-performance computers to process the acquired images. To address these challenges, we propose a novel microarray detection (MD) method designed for deployment on a charge-coupled device (CCD) microarray scanner integrated with an edge computing. In this MD method, we propose a novel model named C2f_SC, which integrates the computational efficiency of the star operation with contextual anchor attention to enhance detection accuracy within a compact model architecture. We further integrate the C2f_SC module into the You Only Look Once v8 (YOLOv8) framework, utilizing MobileNetV3 with the convolutional block attention module (CBAM) as the backbone, removing the detection heads for medium and large objects, replacing the bounding box loss function with the signed intersection over union, and incorporating ResBlock_CBAM before the small object detection head. With the microarray images from the gene expression omnibus database, experimental results demonstrate that our method significantly reduces the required giga floating point operations per seconds (GFLOPs) and model size compared to other YOLO models, while maintaining comparable or higher mAP@50 performance with only a slight drop at mAP@50–95. Specifically, against YOLOv8n, our approach matches mAP@50 while using just 32.9% of GFLOPs and 43.3% of model weights. In addition, we conducted a comparative experiment between the laser confocal scanner and the CCD scanner on the BAC Microarray from MGmed Inc., Republic of Korea (MGMED BAC) to verify the reliability of the CCD scanner. Extensive experimental results demonstrate that the proposed MD method significantly reduces reliance on large-scale computing systems.
- 30 December 2025
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 37 (2) , 026107
A novel microarray detection method based on deep learning with edge computing
Traditional microarray scanners typically rely on high-performance computers to process the acquired images. To address these challenges, we propose a novel microarray detection (MD) method designed for deployment on a charge-coupled device (CCD) microarray scanner integrated with an edge computing. In this MD method, we propose a novel model named C2f_SC, which integrates the computational efficiency of the star operation with contextual anchor attention to enhance detection accuracy within a compact model architecture. We further integrate the C2f_SC module into the You Only Look Once v8 (YOLOv8) framework, utilizing MobileNetV3 with the convolutional block attention module (CBAM) as the backbone, removing the detection heads for medium and large objects, replacing the bounding box loss function with the signed intersection over union, and incorporating ResBlock_CBAM before the small object detection head. With the microarray images from the gene expression omnibus database, experimental results demonstrate that our method significantly reduces the required giga floating point operations per seconds (GFLOPs) and model size compared to other YOLO models, while maintaining comparable or higher mAP@50 performance with only a slight drop at mAP@50–95. Specifically, against YOLOv8n, our approach matches mAP@50 while using just 32.9% of GFLOPs and 43.3% of model weights. In addition, we conducted a comparative experiment between the laser confocal scanner and the CCD scanner on the BAC Microarray from MGmed Inc., Republic of Korea (MGMED BAC) to verify the reliability of the CCD scanner. Extensive experimental results demonstrate that the proposed MD method significantly reduces reliance on large-scale computing systems.
- 17 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (15) , 29711-29721
A Low-Complexity YOLOv8s-Based Model for Environment Perception in Power Maintenance Guided Vehicles
Traditional methods for moving power covers in the power maintenance sector introduce complexities and safety risks due to their heavy reliance on manual labor. In order to reduce dependence on human resources while enhancing safety in operational processes, we introduce a low-complexity model called YOLO-EPGV (Environmental Perception in Guided Vehicles YOLO), which is built upon YOLOv8s to address this challenge. This model is specifically tailored for environmental perception in engineering vehicles used for power maintenance, aiming to advance vehicle automation, and minimize labor requirements and accident risks. YOLO-EPGV combines the StartNet architecture, Partial Convolution (PConv), and Convolutional Gated Linear Units (CGLU) to improve detection precision (mAP@50) while reduce computational costs. Furthermore, we developed a new greenbelt dataset, which was used for training and evaluating the model. The results show that YOLO-EPGV delivers excellent performance, achieving a mAP@50 of 92.8% with minimal computational load, outperforming several YOLO variants. Ablation studies confirm the contribution of each component to the overall performance, and comparisons with other lightweight backbone networks highlight the superior efficiency and precision of StartNet. Subsequently, the model was deployed on the NVIDIA Jetson Orin Nano, achieving a detection speed of 57 FPS while maintaining mAP@50 of 92.7%. This achievement has been validated through empirical testing in real-world operational settings.
- 17 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (15) , 29711-29721
A Low-Complexity YOLOv8s-Based Model for Environment Perception in Power Maintenance Guided Vehicles
Traditional methods for moving power covers in the power maintenance sector introduce complexities and safety risks due to their heavy reliance on manual labor. In order to reduce dependence on human resources while enhancing safety in operational processes, we introduce a low-complexity model called YOLO-EPGV (Environmental Perception in Guided Vehicles YOLO), which is built upon YOLOv8s to address this challenge. This model is specifically tailored for environmental perception in engineering vehicles used for power maintenance, aiming to advance vehicle automation, and minimize labor requirements and accident risks. YOLO-EPGV combines the StartNet architecture, Partial Convolution (PConv), and Convolutional Gated Linear Units (CGLU) to improve detection precision (mAP@50) while reduce computational costs. Furthermore, we developed a new greenbelt dataset, which was used for training and evaluating the model. The results show that YOLO-EPGV delivers excellent performance, achieving a mAP@50 of 92.8% with minimal computational load, outperforming several YOLO variants. Ablation studies confirm the contribution of each component to the overall performance, and comparisons with other lightweight backbone networks highlight the superior efficiency and precision of StartNet. Subsequently, the model was deployed on the NVIDIA Jetson Orin Nano, achieving a detection speed of 57 FPS while maintaining mAP@50 of 92.7%. This achievement has been validated through empirical testing in real-world operational settings.
- 18 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
A Novel Equipment Based on Improved YOLOv5s for Automated Electrical Cover Handling
With the rapid development of the power industry, the number of power cables has increased significantly, making the improvement of maintenance efficiency a critical issue. Reducing accidents caused by the manual handling of heavy electrical cover plates has become a pressing concern. However, the maintenance of power cables often requires precise positioning and efficient movement of these heavy cover plates in energy-poor environments. Existing object detection models typically not only require high-performance computing but also have with high cost and large volume, rendering them quite unsuitable for deployment in the resource-constrained system. To address these challenges, we propose an automated solution that integrates negative pressure adsorption and computer vision techniques. The core module of this method is Cover Plates Edge Line-YOLO (CPEL-YOLO), a lightweight model based on You Only Look Once (YOLO). In CPEL-YOLO, it leverages MobileNetV3 with efficient multiscale attention (EMA) and C3-Faster with EMA to focus the model on precise edge lines with reduced computational complexity. Additionally, Scylla-IoU (SIoU) is incorporated to optimize the training process. A dataset of 1009 cover plate images was constructed for training and evaluation. Experimental results show that CPEL-YOLO achieves a mAP@0.5 of 88.1%, comparable to YOLOv5s, while reducing model size by 42% and improving precision by 4.4%.
- 18 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
A Novel Equipment Based on Improved YOLOv5s for Automated Electrical Cover Handling
With the rapid development of the power industry, the number of power cables has increased significantly, making the improvement of maintenance efficiency a critical issue. Reducing accidents caused by the manual handling of heavy electrical cover plates has become a pressing concern. However, the maintenance of power cables often requires precise positioning and efficient movement of these heavy cover plates in energy-poor environments. Existing object detection models typically not only require high-performance computing but also have with high cost and large volume, rendering them quite unsuitable for deployment in the resource-constrained system. To address these challenges, we propose an automated solution that integrates negative pressure adsorption and computer vision techniques. The core module of this method is Cover Plates Edge Line-YOLO (CPEL-YOLO), a lightweight model based on You Only Look Once (YOLO). In CPEL-YOLO, it leverages MobileNetV3 with efficient multiscale attention (EMA) and C3-Faster with EMA to focus the model on precise edge lines with reduced computational complexity. Additionally, Scylla-IoU (SIoU) is incorporated to optimize the training process. A dataset of 1009 cover plate images was constructed for training and evaluation. Experimental results show that CPEL-YOLO achieves a mAP@0.5 of 88.1%, comparable to YOLOv5s, while reducing model size by 42% and improving precision by 4.4%.
- 4 September 2024
- Published by Elsevier in Expert Systems with Applications
- Vol. 259
SGRN: SEMG-based gesture recognition network with multi-dimensional feature extraction and multi-branch information fusion
- 4 September 2024
- Published by Elsevier in Expert Systems with Applications
- Vol. 259
SGRN: SEMG-based gesture recognition network with multi-dimensional feature extraction and multi-branch information fusion
- Open Access
- 7 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Photonics Journal
- Vol. 16 (4) , 1-11
CCD Standard Curve Fitting for Microarray Detection Base on Multi-Layer Perceptron
The Charge Coupled Device (CCD) scanner determines the concentration of the microarray by capturing the intensity of the fluorescent signal on the microarray in combination with the standard curve. Due to the characteristics of semiconductors, the CCD sensor in the scanner we designed suffers from saturation, the non-linear relationship between photoelectric response and the light intensity collected by CCD, which poses a challenge for fitting the standard curve of microarray scanner. The Least Squares Algorithm (LSA) still has a large relative error even in the case of high-order fitting, especially in the region of the fluorescence image with small gray level. However, the standard curve is critical to the highly accurate measuring of the instrument. In view of the poor curve fitting performance of LSA, Weighted Least Squares (WLS), and Penalized Least Squares (PLS), as well as the small dataset, this paper proposes the Multi-Layer Perceptron (MLP) neural network algorithm with the minimization of relative error as the constraint, which is applied to the standard curve fitting of the scanner. The gray-level of the fluorescent probe in detection image was obtained as the data set acquired by the microarray scanner at different exposure time. And the relative error and the standard deviation of the relative errors were used as evaluation indicators. In our experiments we compared the MLP neural network with relative error minimization as the constraint with the LSA and the MLP neural network with sum of square errors (SSE) minimization as the constraint. The experimental results show that the MLP neural network constrained by minimizing the relative error has good fitting performance for the standard curve of CCD scanner, with the maximum relative error of only 0.89% while the standard deviation of relative error of only 0.25%. It can be seen that this method provides a new approach for standard curve fitting of microarray scanner.
- 7 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Photonics Journal
- Vol. 16 (4) , 1-11
Open AccessCCD Standard Curve Fitting for Microarray Detection Base on Multi-Layer Perceptron
The Charge Coupled Device (CCD) scanner determines the concentration of the microarray by capturing the intensity of the fluorescent signal on the microarray in combination with the standard curve. Due to the characteristics of semiconductors, the CCD sensor in the scanner we designed suffers from saturation, the non-linear relationship between photoelectric response and the light intensity collected by CCD, which poses a challenge for fitting the standard curve of microarray scanner. The Least Squares Algorithm (LSA) still has a large relative error even in the case of high-order fitting, especially in the region of the fluorescence image with small gray level. However, the standard curve is critical to the highly accurate measuring of the instrument. In view of the poor curve fitting performance of LSA, Weighted Least Squares (WLS), and Penalized Least Squares (PLS), as well as the small dataset, this paper proposes the Multi-Layer Perceptron (MLP) neural network algorithm with the minimization of relative error as the constraint, which is applied to the standard curve fitting of the scanner. The gray-level of the fluorescent probe in detection image was obtained as the data set acquired by the microarray scanner at different exposure time. And the relative error and the standard deviation of the relative errors were used as evaluation indicators. In our experiments we compared the MLP neural network with relative error minimization as the constraint with the LSA and the MLP neural network with sum of square errors (SSE) minimization as the constraint. The experimental results show that the MLP neural network constrained by minimizing the relative error has good fitting performance for the standard curve of CCD scanner, with the maximum relative error of only 0.89% while the standard deviation of relative error of only 0.25%. It can be seen that this method provides a new approach for standard curve fitting of microarray scanner.
- 16 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 24 (13) , 21595-21603
Standard Curve Fitting of CCD Fluorescence Scanner Based on Neural Dynamics
The substantial relative error encountered in fitting the standard curve of Charge Coupled Device (CCD) measurements, particularly in microarray probes with low responses, significantly undermines the overall accuracy. To mitigate this challenge, we introduce a novel approach that leverages the Neural Dynamics Optimization Algorithm (NDOA) to address the constraint of minimizing relative error. This method demonstrates superior performance compared to the Least Squares Algorithm (LSA), effectively overcoming the limitations posed by the larger relative errors associated with LSA. The experimental results, including comparative fitting from all probes and linear region’s probes, reveal that the new NDOA-optimized fitting method has good performance for the standard curve of CCD scanner. We could draw a conclusion that there are notable fitting advantages by NDOA which presents the superior fitting accuracy compared to LSA. Meanwhile, the fitting performance in linear region is better than that of responds in all probes. Additionally, the maximum relative error decreases to 0.53% @ N=3 with NDOA, particularly in linear region. Especially, the simple linear fitting (N=1) method excluding saturated probes exhibits more excellent accuracy, decreases to 3.24% @ N=1, than that of LSA. These exceptional metrics indicate that this newly method demonstrates excellent capability in fitting the standard curve of CCD scanner so as to improve the detection precision for microarray.
- 16 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 24 (13) , 21595-21603
Standard Curve Fitting of CCD Fluorescence Scanner Based on Neural Dynamics
The substantial relative error encountered in fitting the standard curve of Charge Coupled Device (CCD) measurements, particularly in microarray probes with low responses, significantly undermines the overall accuracy. To mitigate this challenge, we introduce a novel approach that leverages the Neural Dynamics Optimization Algorithm (NDOA) to address the constraint of minimizing relative error. This method demonstrates superior performance compared to the Least Squares Algorithm (LSA), effectively overcoming the limitations posed by the larger relative errors associated with LSA. The experimental results, including comparative fitting from all probes and linear region’s probes, reveal that the new NDOA-optimized fitting method has good performance for the standard curve of CCD scanner. We could draw a conclusion that there are notable fitting advantages by NDOA which presents the superior fitting accuracy compared to LSA. Meanwhile, the fitting performance in linear region is better than that of responds in all probes. Additionally, the maximum relative error decreases to 0.53% @ N=3 with NDOA, particularly in linear region. Especially, the simple linear fitting (N=1) method excluding saturated probes exhibits more excellent accuracy, decreases to 3.24% @ N=1, than that of LSA. These exceptional metrics indicate that this newly method demonstrates excellent capability in fitting the standard curve of CCD scanner so as to improve the detection precision for microarray.