计算机应用研究2026,Vol.43Issue(2):332-341,10.DOI:10.19734/j.issn.1001-3695.2025.05.0230
基于深度学习的小样本图像分类方法综述
Survey of few-shot image classification based on deep learning
摘要
Abstract
This paper addressed the challenges posed by limited training resources and stringent deployment requirements for focusing on few-shot image classification methods,which enabled accurate model training with minimal labeled data.This paper conducted a systematic review of deep learning-based few-shot classification algorithms,outlining their general pipeline and evaluation metrics.It categorized existing approaches into optimization-based,metric-based,and multimodal-based groups,summarized the key technical routes and representative algorithms for each category.Furthermore,it analyzed performance trends of existing methods on few-shot datasets,highlighting future research.关键词
深度学习/人工智能/小样本学习/图像分类/多模态/大语言模型Key words
deep learning/artificial intelligence/few-shot learning(FSL)/image classification/multi-modal/large language model分类
信息技术与安全科学引用本文复制引用
陈镜宇,吴加莹,罗佳,郁文倩,古月敬之,仲兆满..基于深度学习的小样本图像分类方法综述[J].计算机应用研究,2026,43(2):332-341,10.基金项目
国家自然科学基金资助项目(72174079) (72174079)
北京市自然科学基金资助项目(9242003) (9242003)
江苏省高等学校自然科学研究项目(23KJB520007) (23KJB520007)
重庆市自然科学基金资助项目(CSTB2023NSCQ-MSX0391) (CSTB2023NSCQ-MSX0391)
江苏省"青蓝工程"大数据优秀教学团队资助项目(2022-29) (2022-29)
连云港市重点研发计划(产业前瞻与关键核心技术)资助项目(CG2323) (产业前瞻与关键核心技术)