智能系统学报2024，Vol.19Issue(6)：1385-1394,10.DOI:10.11992/tis.202310025
ObjectBoxG:基于GC3模块的目标检测算法
ObjectBoxG:object detection algorithm based on GC3 module
摘要
Abstract
With the deepening development of the study on object detection tasks,anchor-free methods such as the Ob-jectBox detector have attracted the attention of researchers.However,the ObjectBox detector has its limitations:it does not fully utilize multiscale features or adequately consider the correlation between target center points and global in-formation.A graph convolution layer module(GConv),which is based on the graph spectrum method,is proposed to learn global image features and address the aforementioned limitations.Additionally,a new module named GC3 com-bines the proposed GConv module with C3(cross-stage partial network with 3 conversions)to further extract the origin-al,fine,and global image features.GC3 is combined with the generalized feature pyramid network(GGFPN)to form the GGFPN.The GGFPN is then embedded into the ObjectBox detector,resulting in the ObjectBoxG algorithm.Experi-ments on benchmark datasets demonstrate that the proposed GC3 module has stronger feature extraction capability than the original C3 module,and the proposed GGFPN network offers superior feature learning capability to GC3.The Ob-jectBoxG algorithm demonstrates excellent performance in object detection.关键词
图卷积神经网络/特征提取/特征融合/目标检测/深度学习/无锚框方法/特征金字塔网络/Object-Box检测器/多尺度特征/全局特征Key words
graph convolutional neural network/feature extraction/feature fusion/object detection/deep learning/an-chor-freem ethods/feature pyram id network/Object-Box detector/multi-scale features/global features分类
信息技术与安全科学引用本文复制引用
张建宇,谢娟英..ObjectBoxG:基于GC3模块的目标检测算法[J].智能系统学报,2024,19(6):1385-1394,10.基金项目
国家自然科学基金项目(62076159,61673251,12031010) （62076159,61673251,12031010）
中央高校基本科研业务费项目(GK202105003). （GK202105003）
