基于改進(jìn)YOL0v8+DeepS0RT的多目標(biāo)車輛跟蹤算法研究

打開文本圖片集

關(guān)鍵詞：車輛目標(biāo)跟蹤；YOLOv8；DeepSORT；Ghost卷積；輕量型；CBAM；損失函數(shù)中圖分類號(hào)：TP391.4；TP301.6 文獻(xiàn)標(biāo)識(shí)碼：A 文章編號(hào)：2096-4706（2025）07-0052-06

Abstract：Aiming at the issues that vehicle tracking algorithms are highlycomplex and computationally intensive in practical traficscenarios，making it diffcult toapplythem todevices withlimited resources，this paper proposesamultitargetvehicletrackingalgorithmbasedonimproved YOLOv8 + DeepSORT.AlightweightGhostNet ConvolutionalNeural Network is introduced intothe backbone network，and the Conv is replaced with GhostConv.This replacement not only ensures the lightweight natureof themodel but also improves its performance.Subsequently，by introducing the CBAMand integrating it with the Ghost convolution technology，anew GC-C2ffeature fusion module is constructed to further enhance the featureextractionability.Finallyanewlossunction，WIoU，isadoptedtoimprovethmodel'segreionacacyand convergence speed.The detectionresults ofthe improved YOLOv8 modelare usedasthe input ofthe DepSORTalgorithm toachievemulti-targetvehicletracking incomplexsituations.Experimentalresultsdemonstratethaton theKITTItraffic dataset， withoutsacrificing detection accuracy，compared with theoriginal YOLOv8+DeepSORT，the parametercountof GCWYOLO+DeepSORT is reduced by 3 5 . 9 4 % and the computational load is decreased by 2 0 . 2 5 % .This makes it more suitable for deployment on devices with limited resources and endows it with practical value.

KeyWords：vehicle target tracking; YOLOv8; DeepSORT; Ghost convolution; Lightweight type; CBAM; Loss Functic

0 引言

在國(guó)家政策的引導(dǎo)和大力支持下，我國(guó)計(jì)算機(jī)視覺技術(shù)迎來(lái)了快速發(fā)展的黃金時(shí)期。（剩余8692字）