宿州学院学报2025，Vol.40Issue(3)：6-10,17,6.DOI:10.3969/j.issn.1673-2006.2025.03.002
基于YOLOv7模型应用于小目标检测算法的改进策略研究
Research on the Improvement Strategy of Small Target Detection Algorithm Based on YOLOv7 Model
摘要
Abstract
Based on the YOLOv7 target detection algorithm,a convolutional neural network structure integrating spa-tially separable convolutions is designed to improve the detection performance of the algorithm on small target data-sets.First,the ECA attention mechanism module is introduced in the Neck part of YOLOv7,which facilitates the ac-curate target localization and feature extraction by the backbone network by weighting the importance of the feature map.Second,to solve the problem of low resolution of small target detection,the spatially separable convolutional Spd-Conv in the CNN module is employed for processing low-resolution and small-object images,and a tiny-target detection layer is added to enhance the detection of small targets.Finally,the GIOU loss function is introduced to further improve the detection on small targets.Experimental results show that the proposed method achieves signifi-cant detection results on the Tiny Person dataset,verifying its effectiveness and applicability.关键词
YOLOv7/深度学习/计算机视觉/小目标检测Key words
YOLOv7/deep learning/computer vision/small target detection分类
信息技术与安全科学引用本文复制引用
卞光权,徐勤军,刘明君,何进成..基于YOLOv7模型应用于小目标检测算法的改进策略研究[J].宿州学院学报,2025,40(3):6-10,17,6.基金项目
安徽省高校自然科学研究重点项目(KJ2021A0657) （KJ2021A0657）
阜阳师范大学博士启动项目(2021KYQD0036). （2021KYQD0036）
