文献类型：期刊文献

英文题名：Efficient Detection Framework Adaptation for Edge Computing: A Plug-and-play Neural Network Toolbox Enabling Edge Deployment

作者：Wu, Jiaqi[1,2,3]; Zhang, Shihao[2]; Chen, Simin[4]; Wang, Lixu[5]; Wang, Zehua[3]; Chen, Wei[6]; He, Fangyuan[7]; Tian, Zijian[2]; Yu, F. Richard[8]; Leung, Victor C.M.[3,9,10]

第一作者：Wu, Jiaqi

机构：[1] Tsinghua University, Department of Automation, Beijing, 100083, China; [2] China University of Mining and Technology [Beijing], School of Artificial Intelligence, Beijing, 100083, China; [3] University of British Columbia, Department of Electrical and Computer Engineering, Vancouver, BC, V6T 1Z4, Canada; [4] university of Texas at dallas, Richardson, TX, 75080, United States; [5] Northwestern University, Evanston, 60208, United States; [6] China University of Mining and Technology, School of Computer Science and Technology, Xuzhou, 221116, China; [7] Beijing Union University, College of Applied Science and Technology, Beijing, 100012, China; [8] Carleton University, Department of Systems and Computer Engineering, Ottawa, ON, K1S 5B6, Canada; [9] Shenzhen MSU-BIT University, Artificial Intelligence Research Institute, Shenzhen, 518172, China; [10] Shenzhen University, College of Computer Science and Software Engineering, Shenzhen, 528060, China

第一机构：Tsinghua University, Department of Automation, Beijing, 100083, China

年份：2025

外文期刊名：IEEE Transactions on Mobile Computing

收录：EI(收录号：20254019262177)

语种：英文

外文关键词：Accident prevention - Arts computing - Convolution - Convolutional neural networks - Deep learning - Edge detection - Human computer interaction - Object detection - Object recognition - Safety devices - Security systems

摘要：Recently, edge computing has emerged as a prevailing paradigm in applying deep learning-based object detection models, offering a promising solution for time-sensitive tasks. However, existing edge object detection faces several challenges: 1) These methods struggle to balance detection precision and model lightweightness. 2) Existing generalized edge-deployment designs offer limited adaptability for object detection. 3) Current works lack real-world evaluation and validation. To address these challenges, we propose the Edge Detection Toolbox (ED-TOOLBOX), which leverages generalizable plug-and-play components to enable edge-site adaptation of object detection models. Specifically, we propose a lightweight Reparameterized Dynamic Convolutional Network (Rep-DConvNet) that employs a weighted multi-shape convolutional branch structure to enhance detection performance. Furthermore, ED-TOOLBOX includes a Sparse Cross-Attention (SC-A) network that adopts a localized-mapping-assisted self-attention mechanism to facilitate a well-crafted Joint Module in adaptively transferring features for further performance improvement. Moreover, we propose an Efficient Head for the classification and location modules to achieve more efficient prediction. Additionally, in practical industrial scenarios, we identify that helmet detection-one of the most representative edge object detection tasks-overlooks band fastening, which introduces potential safety hazards. To address this, we build a Helmet Band Detection Dataset (HBDD) and apply an edge object detection model optimized by the ED-TOOLBOX to tackle this real-world task. Extensive experiments validate the effectiveness of components in ED-TOOLBOX. In visual surveillance simulations, ED-TOOLBOX-assisted edge detection models outperform six state-of-the-art methods, enabling real-time and accurate detection. These results demonstrate that our approach offers a superior solution for edge object detection. ? 2002-2012 IEEE.