基于车载视觉传感器的多目标检测跟踪技术研究  

作　　者：鲜茂森 何季芮 

机构地区：[1]巴中职业技术学院,四川巴中636000

出　　处：《科技创新与应用》2025年第2期110-113,117,共5页Technology Innovation and Application

摘　　要：汽车在现代社会生活中扮演着举足轻重的角色,其极大地便利人们的出行,但与此同时,也引发诸如环境污染和交通拥堵等社会问题。为应对这些挑战,国内外学者正积极投身于无人驾驶技术的研究。其中,车辆与行人检测技术是无人驾驶技术的关键环节,但鉴于车辆与行人特征的多样性,使用单一的特征提取和分类方法进行检测变得异常困难。然而,深度学习的目标检测方案凭借其复杂的神经网络结构,成功攻克这一难题,因此备受国内外学者的关注。该文旨在通过深入剖析基于深度学习的目标检测以及基于SORT的多目标跟踪原理,为车载视觉传感器的研发创新提供有益的参考和启示。Automobiles play a pivotal role in modern social life.They greatly facilitate people's travel,but at the same time,they also cause social problems such as environmental pollution and traffic congestion.In order to meet these challenges,domestic and foreign scholars are actively engaged in the research of driverless technology.Among them,vehicle and pedestrian detection technology is a key link in driverless technology.However,given the diversity of vehicle and pedestrian characteristics,it has become extremely difficult to detect using a single feature extraction and classification method.However,the deep learning target detection scheme has successfully overcome this problem with its complex neural network structure,so it has attracted the attention of scholars at home and abroad.This paper aims to provide useful reference and inspiration for the research and development and innovation of vehicle-mounted vision sensors by in-depth analysis of target detection based on deep learning and multi-target tracking based on SORT.

关 键 词：车载视觉传感器 多目标检测 跟踪技术 无人驾驶 车辆与行人检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]