基于小目标检测的YOLO算法研究综述  

Review of YOLO Algorithm Based on Small Target Detection

作　　者：杨成佳[1] 钱明 YANG Cheng-jia;QIAN Ming(Jilin Jianzhu University,Changchun Jilin 130118,China)

机构地区：[1]吉林建筑大学,吉林长春130118

出　　处：《吉林工程技术师范学院学报》2023年第3期92-96,共5页Journal of Jilin Engineering Normal University

基　　金：吉林省教育厅科学技术研究项目(JJKH20220276KJ)。

摘　　要：小目标检测是目标检测中的一个难点,小目标检测需要识别图片中像素少,分辨率、特征信息少的目标的位置和种类,目前应用成熟的中、大目标检测使用的算法在检测小目标时效果不理想。所以,提高小目标检测的能力是当前目标检测领域的难题,也是重要的研究方向。本文先对小目标检测的定义和YOLO算法的发展优化进行介绍,介绍分析YOLO算法在实际场景应用对小目标检测的应用优化,并对优化方案进行汇总概括,最后对未来YOLO算法在小目标识别中的发展趋势进行预测。Small target detection is a difficult point in target detection because it needs to identify the location and types of targets with few pixels,low resolution and less feature information in the picture.At present,the mature algorithms used in medium and large target detection are not ideal in detecting small targets.Therefore,improving the ability of small target detection is a difficult problem in the field of target detection and an important research focus.In this paper,the definition of small target detection and the development optimization of YOLO algorithm are introduced,and the application optimization of YOLO algorithm in actual scene is analyzed,followed by a summary of optimization schemes.Finally,the development trend of YOLO algorithm in small target recognition in the future is predicted.

关 键 词：小目标检测 YOLO算法 应用与优化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]