Query：
学者姓名：林志贤
Refining：
Year
Type
Indexed by
Source
Complex
Former Name
Co-
Language
Clean All
Abstract ：
针对当前三维目标检测由于数据增强导致点云和图像无法有效对齐,点与点对齐方法会丢失图像特征以及定位和分类置信度不一致的问题,提出一种多模态融合的三维目标检测方法.首先,采用PointNet++提取点云的特征;采用卷积神经网络提取图像特征;其次,在点云与图像融合阶段,采用语义对齐方法以及图像球特征,实现点云与图像更好的跨模态对齐.同时采用基于注意力的方法来指导点云与图像特征的融合,以获取更可靠的图像特征;最后引入DIoU损失来平衡置信度不一致的问题.实验结果表明:所采用的方法明显优于baseline,在简单、中等和困难任务下,Car类别的mAP达85.6%.
Keyword ：
多模态融合 多模态融合 彩色图像 彩色图像 激光雷达 激光雷达 自动驾驶 自动驾驶
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | 韩路宇 , 林珊玲 , 赵民 et al. 基于球语义多模态融合的三维目标检测 [J]. | 光电子技术 , 2025 , 45 (1) : 75-81 . |
| MLA | 韩路宇 et al. "基于球语义多模态融合的三维目标检测" . | 光电子技术 45 . 1 (2025) : 75-81 . |
| APA | 韩路宇 , 林珊玲 , 赵民 , 林志贤 , 郭太良 . 基于球语义多模态融合的三维目标检测 . | 光电子技术 , 2025 , 45 (1) , 75-81 . |
| Export to | NoteExpress RIS BibTex |
Version ：
Abstract ：
针对目前驾驶员疲劳检测算法存在检测过程复杂、参数多、精度低、运行速度慢等问题,提出了一种基于改进YOLOv8n-Pose的轻量级模型.该模型优化了YOLOv8n-Pose的结构.首先,在模型主干网络中,引入Ghost卷积减少模型参数量和不必要的卷积计算.其次,引入Slim-neck融合主干网络提取的不同尺寸特征,加速网络预测计算.同时在颈部网络添加遮挡感知注意力模块(SEAM),强调图像中的人脸区域并弱化背景,改善关键点定位效果.最后,在检测头部分提出一种GNSC-Head结构,引入共享卷积,并将传统卷积的BN层优化成更稳定的GN层,有效节省模型的参数空间和计算资源.实验结果显示,改进后的YOLOv8n-Pose相较于原始算法,mAP@0.5提高了0.9%,参数量和计算量各减少了50%,同时FPS提高了8%,最终的疲劳驾驶识别率达到93.5%.经验证,本文算法在轻量化的同时能够保持较高的检测精度,并且能够有效识别驾驶员状态,为车辆边缘设备的部署提供有力支撑.
Keyword ：
YOLOv8n-Pose YOLOv8n-Pose 注意力机制 注意力机制 深度学习 深度学习 疲劳驾驶检测 疲劳驾驶检测 轻量化 轻量化
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | 蔡忠祺 , 林珊玲 , 林坚普 et al. 基于改进YOLOv8n-Pose的疲劳驾驶检测 [J]. | 液晶与显示 , 2025 , 40 (4) : 617-629 . |
| MLA | 蔡忠祺 et al. "基于改进YOLOv8n-Pose的疲劳驾驶检测" . | 液晶与显示 40 . 4 (2025) : 617-629 . |
| APA | 蔡忠祺 , 林珊玲 , 林坚普 , 吕珊红 , 林志贤 , 郭太良 . 基于改进YOLOv8n-Pose的疲劳驾驶检测 . | 液晶与显示 , 2025 , 40 (4) , 617-629 . |
| Export to | NoteExpress RIS BibTex |
Version ：
Abstract ：
电润湿电子纸采用减色混色系统进行色彩显示,色域较小,容易发生色彩失真,且依赖环境光的漫反射,亮度不足.为解决这些问题,提出一种基于彩色电润湿的色彩空间转换和图像自适应增强算法.该算法将图像从RGB色彩空间转换到HSV空间,并使用CLAHE对饱和度进行均匀分布处理,改善色彩表现.亮度通道通过引导滤波和改进的Retinex算法进行增强,保留细节与边缘信息,使电润湿电子纸在相同光照下依旧保持真实视觉效果.实验结果表明,该算法在PSNR、SSIM、FSIM和FSIMc上分别提高了19%、10.8%、19.19%和19.54%,显著优化电润湿电子纸的显示效果,为其市场化应用提供有力支撑.
Keyword ：
图像增强 图像增强 彩色电润湿电子纸 彩色电润湿电子纸 直方图均衡 直方图均衡 色彩空间变换 色彩空间变换
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | 毛文杰 , 林珊玲 , 林坚普 et al. 基于色彩空间变换的电润湿电子纸色彩校正 [J]. | 光电工程 , 2025 , 52 (2) : 32-45 . |
| MLA | 毛文杰 et al. "基于色彩空间变换的电润湿电子纸色彩校正" . | 光电工程 52 . 2 (2025) : 32-45 . |
| APA | 毛文杰 , 林珊玲 , 林坚普 , 梅婷 , 王廷雨 , 蔡苾芃 et al. 基于色彩空间变换的电润湿电子纸色彩校正 . | 光电工程 , 2025 , 52 (2) , 32-45 . |
| Export to | NoteExpress RIS BibTex |
Version ：
Abstract ：
针对野生动物数据集样本量小、目标尺度多变所导致的野生动物检测困难以及检测精度低等问题,提出一种基于多尺度上下文提取的小样本野生动物检测(MS-FSWD)算法.首先,通过多尺度上下文提取模块增强模型对不同尺度的野生动物的感知能力,提高检测性能;其次,引入Res2Net作为原型校准模块的强分类网络对分类器输出的分类分数进行校正;然后,在RPN中加入置换注意力机制,增强目标区域的特征图,弱化背景信息;最后,将平衡L1损失作为定位损失函数,提升目标定位性能.实验结果表明,相比DeFRCN算法,MS-FSWD在小样本野生动物数据集FSWA上,1-shot和3-shot检测任务中新类AP50分别提升了9.9%和6.6%;在公共数据集PASCAL VOC上,MS-FSWD最高提升了12.6%.与VFA算法相比,在PASCAL VOC数据集Novel Set 3的10-shot任务中,新类AP50提升了3.3%.
Keyword ：
多尺度上下文提取 多尺度上下文提取 小样本目标检测 小样本目标检测 注意力机制 注意力机制 迁移学习 迁移学习 野生动物检测 野生动物检测
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | 刘珂 , 林珊玲 , 师欣雨 et al. 基于多尺度上下文提取的小样本野生动物检测 [J]. | 液晶与显示 , 2025 , 40 (3) : 516-526 . |
| MLA | 刘珂 et al. "基于多尺度上下文提取的小样本野生动物检测" . | 液晶与显示 40 . 3 (2025) : 516-526 . |
| APA | 刘珂 , 林珊玲 , 师欣雨 , 林坚普 , 吕珊红 , 林志贤 et al. 基于多尺度上下文提取的小样本野生动物检测 . | 液晶与显示 , 2025 , 40 (3) , 516-526 . |
| Export to | NoteExpress RIS BibTex |
Version ：
Abstract ：
Due to the problem of hysteretic contact angle and inconsistent aperture caused by charge capture in chromatic inks of electrowetting electronic paper, the reflected brightness and contrast of the display are reduced. The existing optimization methods of active luminous display do not consider the properties of reflective passive display. Combined with chromatic electrowetting characteristics, an electrowetting display algorithm of the brightness enhancement is proposed based on dynamic histogram equalization and multi-scale gamma correction. We perform histogram equalization correction on the brightness by combining the weighting of the aperture ratio and the distribution of the illumination components. Then, corresponding compensation weights are designed based on the different reflection brightness. Furthermore, the illumination component is extracted by multi-scale Gaussian convolution, and multi-scale gamma correction based on different photoelectric characteristics is designed. Ultimately, through threshold discrimination, the images of enhanced brightness are outputted to compensate for the missing or insufficient of the display brightness. The experimental results demonstrate that the images processed by this algorithm after brightness enhancement show certain improvements in objective evaluation indicators. More importantly, the images processed by this algorithm exhibit better brightness details on chromatic electrowetting displays.
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Chen, Mingzhen , Lin, Zhixian , Lin, Shanling et al. Electrowetting display of multiscale Gamma based on dynamic histogram equilibrium [J]. | SCIENTIFIC REPORTS , 2025 , 15 (1) . |
| MLA | Chen, Mingzhen et al. "Electrowetting display of multiscale Gamma based on dynamic histogram equilibrium" . | SCIENTIFIC REPORTS 15 . 1 (2025) . |
| APA | Chen, Mingzhen , Lin, Zhixian , Lin, Shanling , Lin, Jianpu , Guo, Tailiang . Electrowetting display of multiscale Gamma based on dynamic histogram equilibrium . | SCIENTIFIC REPORTS , 2025 , 15 (1) . |
| Export to | NoteExpress RIS BibTex |
Version ：
Abstract ：
With the continuous advancement of embedded system technology, embedded deployment of deep learning models has become a hot research issue. Resource-constrained devices such as low-performance microcontrollers have stringent computational and memory requirements, so how to design efficient and lightweight convolutional neural networks (CNNs)to adapt to the limitations of these devices has become a key research direction. In this paper, we propose a lightweight CNN design scheme based on a simplified LeNet-5 [1] architecture, aiming at efficient handwritten digit recognition through integer weight quantization and pure C programming. The network is trained and tested on the MNIST dataset, and the optimized model is successfully deployed on a low-performance microcontroller and shows significant advantages in terms of computational complexity and memory footprint. In addition, Ibuilt and implemented this lightweight neural network using pure logic gate circuits. Experimental results show that the designed lightweight CNN network has fewer parameters (1326),in addition, the model in this paper has a lower demand in terms of the number of floating point operations (920 FLOPs). It isable to achieve fast processing and significantly reduce the hardware burden while maintaining reasonable accuracy, providing an effective solution for deep learning applications in resource-constrained environments. © 2025, John Wiley and Sons Inc. All rights reserved.
Keyword ：
Computer circuits Computer circuits Convolution Convolution Convolutional neural networks Convolutional neural networks C (programming language) C (programming language) Deep learning Deep learning Digital arithmetic Digital arithmetic Embedded systems Embedded systems Integer programming Integer programming Learning systems Learning systems Logic circuits Logic circuits Logic design Logic design Logic gates Logic gates Memory architecture Memory architecture Microcontrollers Microcontrollers
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Chen, Rixin , Lin, Zhixian , Lin, Jianpu . Lightweight CNN Design and Implementation for Handwritten Digit Recognition on Resource Constrained Devices [C] . 2025 : 909-913 . |
| MLA | Chen, Rixin et al. "Lightweight CNN Design and Implementation for Handwritten Digit Recognition on Resource Constrained Devices" . (2025) : 909-913 . |
| APA | Chen, Rixin , Lin, Zhixian , Lin, Jianpu . Lightweight CNN Design and Implementation for Handwritten Digit Recognition on Resource Constrained Devices . (2025) : 909-913 . |
| Export to | NoteExpress RIS BibTex |
Version ：
Abstract ：
Ovarian cancer stands out as one of the most formidable adversaries in women's health, largely due to its typically subtle and nonspecific early symptoms, which pose significant challenges to early detection and diagnosis. Although existing diagnostic methods, such as biomarker testing and imaging, can help with early diagnosis to some extent, these methods still have limitations in sensitivity and accuracy, often leading to misdiagnosis or missed diagnosis. Ovarian cancer's high heterogeneity and complexity increase diagnostic challenges, especially in disease progression prediction and patient classification. Machine learning (ML) has outperformed traditional methods in cancer detection by processing large datasets to identify patterns missed by conventional techniques. However, existing AI models still struggle with accuracy in handling imbalanced and high-dimensional data, and their "black-box" nature limits clinical interpretability. To address these issues, this study proposes SHAP-GAN, an innovative diagnostic model for ovarian cancer that integrates Shapley Additive exPlanations (SHAP) with Generative Adversarial Networks (GANs). The SHAP module quantifies each biomarker's contribution to the diagnosis, while the GAN component optimizes medical data generation. This approach tackles three key challenges in medical diagnosis: data scarcity, model interpretability, and diagnostic accuracy. Results show that SHAP-GAN outperforms traditional methods in sensitivity, accuracy, and interpretability, particularly with high-dimensional and imbalanced ovarian cancer datasets. The top three influential features identified are PRR11, CIAO1, and SMPD3, which exhibit wide SHAP value distributions, highlighting their significant impact on model predictions. The SHAP-GAN network has demonstrated an impressive accuracy rate of 99.34% on the ovarian cancer dataset, significantly outperforming baseline algorithms, including Support Vector Machines (SVM), Logistic Regression (LR), and XGBoost. Specifically, SVM achieved an accuracy of 72.78%, LR achieved 86.09%, and XGBoost achieved 96.69%. These results highlight the superior performance of SHAP-GAN in handling high-dimensional and imbalanced datasets. Furthermore, SHAP-GAN significantly alleviates the challenges associated with intricate genetic data analysis, empowering medical professionals to tailor personalized treatment strategies for individual patients.
Keyword ：
extreme gradient boosting algorithm extreme gradient boosting algorithm feature selection feature selection generative adversarial networks generative adversarial networks ovarian cancer ovarian cancer SHAP SHAP
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Cai, Jingxun , Lee, Zne-Jung , Lin, Zhihxian et al. A Novel SHAP-GAN Network for Interpretable Ovarian Cancer Diagnosis [J]. | MATHEMATICS , 2025 , 13 (5) . |
| MLA | Cai, Jingxun et al. "A Novel SHAP-GAN Network for Interpretable Ovarian Cancer Diagnosis" . | MATHEMATICS 13 . 5 (2025) . |
| APA | Cai, Jingxun , Lee, Zne-Jung , Lin, Zhihxian , Yang, Ming-Ren . A Novel SHAP-GAN Network for Interpretable Ovarian Cancer Diagnosis . | MATHEMATICS , 2025 , 13 (5) . |
| Export to | NoteExpress RIS BibTex |
Version ：
Abstract ：
Inadequate exposure of imaging devices in low-light environments results in a loss of image information, significantly deteriorating the image quality. However, current low-light image enhancement algorithms commonly suffer from issues such as color distortion and loss of fine details and textures. In this paper, we propose a frequency-guided dual-collapse Transformer (FDCFormer) network. First, in response to color distortion after enhancement, we propose a dual-collapse Transformer that effectively aggregates features from both spatial and channel dimensions, thus capturing global information. Subsequently, because relying solely on enhancement in the spatial domain often makes it difficult to preserve fine details and textures, we design multiple mixed residual fast Fourier transform blocks as additional frequency information guidance branches, focusing on local detail information at the image edges. Additionally, we employ an adaptive dual-domain information fusion module that combines spatial domain and frequency domain information to enrich the final output features. Extensive experiments on multiple publicly available datasets demonstrate that our FDCFormer outperforms state-of-the-art methods, exceeding Retinexformer by up to 0.93 dB on average across five paired datasets. We also employ our method as a preprocessing step in dark detection, our method improves mean average precision (mAP) by 1.9% over the baseline model on ExDark dataset, revealing the latent practical values of our method. The corresponding codes will be available at https://github.com/Fly175/FDCFormer.
Keyword ：
Dual-domain fusion Dual-domain fusion Fourier frequency information Fourier frequency information Low-light image enhancement Low-light image enhancement Vision transformer Vision transformer
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Lin, Jianpu , Lai, Fangwei , Lin, Shanling et al. Frequency-guided dual-collapse Transformer for low-light image enhancement [J]. | ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE , 2025 , 142 . |
| MLA | Lin, Jianpu et al. "Frequency-guided dual-collapse Transformer for low-light image enhancement" . | ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE 142 (2025) . |
| APA | Lin, Jianpu , Lai, Fangwei , Lin, Shanling , Lin, Zhixian , Guo, Tailiang . Frequency-guided dual-collapse Transformer for low-light image enhancement . | ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE , 2025 , 142 . |
| Export to | NoteExpress RIS BibTex |
Version ：
Abstract ：
In order to solve the problem of low gray level due to the few driving chips developed based on the photoelectric characteristics of electrowetting display, a driving method based on modulation is proposed to double the gray level of electrowetting display. In this method, the driving waveform based on pulse amplitude modulation and pulse width modulation hybrid modulation is designed, and the gray level-luminance curve of the electrowetting display is measured and analyzed. On this basis, the luminance nonlinear correction is carried out, and the improvement of 64 Gy levels to 127 Gy levels is realized by the principle of human visual persistence phenomenon. The experimental results show that the proposed driving scheme can break through the limitation of the driving chip and realize the multiplication of gray levels, in which 96% gray levels increase steadily with an average luminance difference of 0.07, and at the same time enhance the contrast and improve the display effect.
Keyword ：
driving waveform driving waveform electrowetting display electrowetting display gray scale gray scale hybrid modulation hybrid modulation nonlinearity nonlinearity PAM PAM PWM PWM
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Mei, Ting , Lin, Zhixian , Xie, Ziyu et al. A driving method for gray scale multiplication of electrowetting display based on hybrid modulation [J]. | FRONTIERS IN PHYSICS , 2024 , 12 . |
| MLA | Mei, Ting et al. "A driving method for gray scale multiplication of electrowetting display based on hybrid modulation" . | FRONTIERS IN PHYSICS 12 (2024) . |
| APA | Mei, Ting , Lin, Zhixian , Xie, Ziyu , Lin, Shanling , Cai, Bipeng , Chen, Mingzhen et al. A driving method for gray scale multiplication of electrowetting display based on hybrid modulation . | FRONTIERS IN PHYSICS , 2024 , 12 . |
| Export to | NoteExpress RIS BibTex |
Version ：
Abstract ：
To address the high power consumption associated with image refresh operations in EPDs, this paper proposes a low-power driving waveform that reduces the refresh power of EPDs by lowering the system's peak power. Compared to traditional waveforms, this waveform first activates the particles before erasing them, thus reducing voltage polarity changes. Additionally, it introduces a specific duration of 0 V voltage during the activation phase based on the physical characteristics of the electrophoretic particles to reduce the voltage span. Finally, a particular duration of 0 V voltage is introduced during the erasure phase to minimize the voltage span while ensuring the stability and consistency of the reference gray scale. The experimental results demonstrate that, in standard power tests, the new driving waveform reduces the power fluctuation value by 1.33% and the energy fluctuation value by 37.24% compared to the traditional driving waveform. This reduction in refresh power also mitigates screen flicker and ghosting phenomena.
Keyword ：
driving waveforms driving waveforms electrophoretic electronic paper electrophoretic electronic paper flicker flicker ghosting ghosting low power consumption low power consumption refresh power consumption refresh power consumption
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Lin, Shanling , Zhang, Jianhao , Wei, Jia et al. Low-Power Driving Waveform Design for Improving the Display Effect of Electrophoretic Electronic Paper [J]. | MICROMACHINES , 2024 , 15 (9) . |
| MLA | Lin, Shanling et al. "Low-Power Driving Waveform Design for Improving the Display Effect of Electrophoretic Electronic Paper" . | MICROMACHINES 15 . 9 (2024) . |
| APA | Lin, Shanling , Zhang, Jianhao , Wei, Jia , Xie, Xinxin , Lv, Shanhong , Mei, Ting et al. Low-Power Driving Waveform Design for Improving the Display Effect of Electrophoretic Electronic Paper . | MICROMACHINES , 2024 , 15 (9) . |
| Export to | NoteExpress RIS BibTex |
Version ：
Export
| Results： |
Selected to |
| Format： |
