Few-Shot Object Detection Based on the Transformer and High-Resolution Network

作     者：Dengyong Zhang Huaijian Pu Feng Li Xiangling Ding Victor S.Sheng 

作者机构：Hunan Provincial Key Laboratory of Intelligent Processing of Big Data on TransportationChangsha University of Science and TechnologyChangsha410114China School of Computer and Communication EngineeringChangsha University of Science and TechnologyChangsha410114China School of Computer Science and EngineeringHunan University of Science and TechnologyXiangtan411004China Department of Computer ScienceTexas Tech UniversityLubbock79409TXUSA 

出 版 物：《Computers, Materials & Continua》 (计算机、材料和连续体（英文）)

年 卷 期：2023年第74卷第2期

页      面：3439-3454页

核心收录：

学科分类：08[工学] 080203[工学-机械设计及理论] 0802[工学-机械工程] 

基　　金：the National Natural Science Foundation of China under grant 62172059 and 62072055 Hunan Provincial Natural Science Foundations of China under Grant 2020JJ4626 Scientific Research Fund of Hunan Provincial Education Department of China under Grant 19B004 “Double First-class”International Cooperation and Development Scientific Research Project of Changsha University of Science and Technology under Grant 2018IC25 the Young Teacher Growth Plan Project of Changsha University of Science and Technology under Grant 2019QJCZ076 

主　　题：Object detection few shot object detection transformer high-resolution 

摘      要：Now object detection based on deep learning tries different *** uses fewer data training networks to achieve the effect of large dataset ***,the existing methods usually do not achieve the balance between network parameters and training *** makes the information provided by a small amount of picture data insufficient to optimize model parameters,resulting in unsatisfactory detection *** improve the accuracy of few shot object detection,this paper proposes a network based on the transformer and high-resolution feature extraction(THR).High-resolution feature extractionmaintains the resolution representation of the *** and spatial attention are used to make the network focus on features that are more useful to the *** addition,the recently popular transformer is used to fuse the features of the existing *** compensates for the previous network failure by making full use of existing object *** on the Pascal VOC and MS-COCO datasets prove that the THR network has achieved better results than previous mainstream few shot object detection.