基于YOLOv5的危险区域多目标检测应用与性能分析.pdfVIP

2024年软   件2024

第45卷 第6期SOFTWAREVol. 45, No.6

基金项目论文

基于YOLOv5的

危险区域多目标检测应用与性能分析

李明昊

（中国石油大学，北京102249）

摘 要：为提高危险区域多目标检测的效率和准确性，以YOLOv5算法为基础，深入分析算法原理和优化策略，使用Qt

研发基于单目视觉与目标检测算法YOLOv5的危险区域多目标检测系统，实现了在危险区域中多目标的高效检测。研究中通过

改进模型架构和模型剪枝、量化，对算法YOLOv5进行了优化，以适应危险区域多目标检测对速度和效率的需求。优化后的算

法有显著的性能提升，平均精度达到97.3%，速度提升近10%，为危险区域多目标检测应用提供了有力的技术支持。

关键词：YOLOv5算法；危险区域；多目标检测；单目视觉；Qt技术

中图分类号：TP391.41 文献标识码：A DOI：10.3969/j.issn.1003-6970.2024.06.007

本文著录格式：李明昊.基于YOLOv5的危险区域多目标检测应用与性能分析[J].软件,2024,45(06):019-021+049

ApplicationandPerformanceAnalysisofMultiObjectDetection

inHazardousAreasBasedonYOLOv5

LIMinghao

(ChinaUniversityofPetroleum,Beijing102249)

【Abstract】：In order to improve the efficiency and accuracy of multi-target detection in hazardous areas,

based on the YOLOv5 algorithm, through in-depth analysis of algorithm principles and optimization strategies,

Qt developed a multi-target detection system for hazardous areas based on monocular vision and object detection

algorithm YOLOv5, achieving efficient detection of multiple targets in hazardous areas. In the study, the YOLOv5

algorithm was optimized by improving the model architecture, pruning, and quantization to meet the speed

and efficiency requirements of multi-target detection in hazardous areas. The optimized YOLOv5 algorithm has

significant performance improvement, with an average accuracy of 97.3% and a speed increase of nearly 10%,

providing strong technical support for multi-target detection applications in hazardous areas.

【Keywo