长春工程学院学报（自然科学版）2025，Vol.26Issue(3)：123-128,6.DOI:10.3969/j.issn.1009-8984.2025.03.019
FDPN:基于特征聚焦和特征扩散机制的金字塔网络在多尺度和小目标检测领域的应用
FDPN:Application of Pyramid Network Based on Feature Focusing and Feature Diffusion Mechanism in Multi Scale and Small Object Detection
摘要
Abstract
Multiscale and small object detection is a core challenge in the field of object detection,as the same object can exhibit different sizes at different distances,making it difficult to capture the features of small objects at long distances.A pyramid network(FDPN)based on feature focusing and feature diffusion mechanism was proposed.By using feature focusing blocks and feature diffusion mechanisms,each scale feature can have detailed contextual information,which is more conducive to subsequent target detection and classification.To verify the performance of FDPN,YOLO framework was used as the basic frame-work.The experimental results show that under the YOLO framework,the performance of FDPN is supe-rior to other pyramid networks.关键词
目标检测/特征提取/特征融合Key words
object detection/feature extraction/feature fusion分类
信息技术与安全科学引用本文复制引用
武文杰,丁言,赵佳..FDPN:基于特征聚焦和特征扩散机制的金字塔网络在多尺度和小目标检测领域的应用[J].长春工程学院学报（自然科学版）,2025,26(3):123-128,6.基金项目
吉林省科技发展计划重点研发项目(20240302080GX) （20240302080GX）
