基于YOLO模型的复杂路面落叶检测研究
Journal: Advances in Computer and Autonomous Intelligence Research DOI: 10.12238/acair.v3i1.11873
Abstract
随着城市化进程的加速,城市绿化面积不断扩大,落叶清理成为城市管理中一项重要而繁琐的任务。传统的人工清理方式不仅效率低下,还存在安全隐患,无法满足现代城市管理的需求。因此,开发一种高效、准确的落叶检测算法,对于提高落叶清理效率、降低人力成本具有重要意义。本文提出了一种基于深度学习的复杂路面落叶检测算法,该算法在YOLOv3目标检测模型的基础上进行改进,通过优化聚类算法、改进激活函数以及引入非极大值融合算法等策略,提高了落叶检测的准确性和实时性。实验结果表明,本文提出的算法在复杂路面落叶检测任务上取得了显著的效果。
Keywords
深度学习；YOLOv3；落叶检测；K-means++
Full Text
PDF - Viewed/Downloaded: 0 TimesReferences
[1] Redmon,J.,&Farhadi,A.(2015).You Only Look Once:Unifie dObject Detection with Deep Neural Networks.In Proceedings of the IEEEconference oncomputer vision and pattern recog nition(pp.5258-5266).
[2] angoordeh,S.,Hebert,P.,&Neumann,M.(2020).YOLOv5:AnUpdated Look at Real-Time Object Detection.arXiv preprintarXiv: 2004.10934.
[3] Chen,P.,&coskun,A.(2021).YOLOv5:Improved Real-TimeOb jectDetection.arXiv preprint arXiv:2104.04764.
[4] Chen Li-fang,Lu Guo-jun.A method of generating MNI ST handwritten digit images based on the fusion of VAE and GAN networks. Journal,2022(2).
[5] Aisha Sajjad. Handwritten Digit Recognition Based on Large-scale Deep Convolutional Neural Networks and MNIST Database[D].Beijing University of Technology.
[2] angoordeh,S.,Hebert,P.,&Neumann,M.(2020).YOLOv5:AnUpdated Look at Real-Time Object Detection.arXiv preprintarXiv: 2004.10934.
[3] Chen,P.,&coskun,A.(2021).YOLOv5:Improved Real-TimeOb jectDetection.arXiv preprint arXiv:2104.04764.
[4] Chen Li-fang,Lu Guo-jun.A method of generating MNI ST handwritten digit images based on the fusion of VAE and GAN networks. Journal,2022(2).
[5] Aisha Sajjad. Handwritten Digit Recognition Based on Large-scale Deep Convolutional Neural Networks and MNIST Database[D].Beijing University of Technology.
Copyright © 2025 姚鹏尊, 闫泽川, 高雨欣, 位盼盼, 赵帅豪, 丁天, 董政
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License
