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

2024-12-24

AI Summary

To address the challenges of simultaneously achieving model lightweighting and high accuracy, poor generalization, and insufficient real-world validation in edge computing–based object recognition, this paper proposes ED-TOOLBOX—a plug-and-play neural network toolbox tailored for edge deployment. Methodologically, it introduces (1) a novel reparameterizable dynamic convolutional network (Rep-DConvNet) to enhance feature representation; (2) a sparse cross-attention (SC-A) mechanism augmented with local mapping to improve modeling of critical components (e.g., seatbelt buckles); and (3) the Helmet Band Detection Dataset (HBDD), the first benchmark dataset dedicated to buckle recognition. Integrated within the YOLO framework, ED-TOOLBOX incorporates a lightweight detection head and end-to-end edge optimization techniques. Experimental results demonstrate a 1.8× speedup over six state-of-the-art methods, a 12.7% accuracy gain on HBDD, and millisecond-level inference latency—validating its practical viability for industrial safety applications.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Object Detection & Categorization
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Systems and Infrastructure for Web, Mobile and WoT: Cloud, edge and content delivery systems for the Web

Abstract

Edge computing has emerged as a key paradigm for deploying deep learning-based object detection in time-sensitive scenarios. However, existing edge detection methods face challenges: 1) difficulty balancing detection precision with lightweight models, 2) limited adaptability of generalized deployment designs, and 3) insufficient real-world validation. To address these issues, we propose the Edge Detection Toolbox (ED-TOOLBOX), which utilizes generalizable plug-and-play components to adapt object detection models for edge environments. Specifically, we introduce a lightweight Reparameterized Dynamic Convolutional Network (Rep-DConvNet) featuring weighted multi-shape convolutional branches to enhance detection performance. Additionally, we design a Sparse Cross-Attention (SC-A) network with a localized-mapping-assisted self-attention mechanism, enabling a well-crafted joint module for adaptive feature transfer. For real-world applications, we incorporate an Efficient Head into the YOLO framework to accelerate edge model optimization. To demonstrate practical impact, we identify a gap in helmet detection -- overlooking band fastening, a critical safety factor -- and create the Helmet Band Detection Dataset (HBDD). Using ED-TOOLBOX-optimized models, we address this real-world task. Extensive experiments validate the effectiveness of ED-TOOLBOX, with edge detection models outperforming six state-of-the-art methods in visual surveillance simulations, achieving real-time and accurate performance. These results highlight ED-TOOLBOX as a superior solution for edge object detection.

Problem

• Edge Computing
• Object Detection
• Neural Network Optimization

Innovation

• Edge Computing
• Rep-DConvNet
• Sparse Cross Attention