农业视觉中的低标注学习：半监督、弱监督与自监督方法综述

打开文本图片集

中图分类号：TP391；S126 文献标志码：A 文章编号：1001-411X（2026）03-0369-13

Learning with limited annotations in agricultural vision： A review of semi-supervised， weakly supervised and self-supervised methods

XIAO Deqin，LIU Qian'，PAN Qianyi'，HUANG J ， TAN Zujie' （1 Collge of Mathematics and Informatics，South China Agricultural University/KeyLaboratoryof Smart Agricultural Technology in Tropical South China， Ministry of Agriculture and Rural Afairs， Guangzhou 510642， China; 2 Guangdong Engineering Research Center for Agricultural Big Data， Guangzhou 510642， China）

Abstract： Traditional agricultural production systems exhibit evident limitations in production effciency， resource utilization and environmental sustainability， and are undergoing a gradual transition toward informatization and intellgent modernization. As a core enabling technology of smart agriculture， agricultural vision plays a crucial role in key applications such as crop production monitoring as well as livestock and poultry breeding management，and has significant practical value for improving agricultural productivity. However，most existing vision modelsrely on large-scale labeled datasets. In agricultural scenarios，complex environments and highly variable data acquisition conditions lead to high annotation costs and long data preparation cycles， which substantially limit large-scale deployment. This paper focuses on three representative learning paradigms with limited annotations in agricultural vision， namely semi-supervised， weakly supervised and self-supervised learning. Their fundamental principles and commonly adopted frameworks are reviewed. The performance characteristics and applicability of these three learning paradigms are summarized in the context of typical agricultural vision tasks. Key challenges， including limited cross-domain generalization and interference from noisy annotations，are further analyzed.Future research directions are discussed， such as the construction of datasets and evaluation benchmarks， the development of agriculture-specific pre-trained models， and active learning-driven low-cost iterative strategies，providing references for the advancement and application of agricultural vision technologies.

KeyWords： Agricultural vision; Smart agriculture; Learning with limited annotations; Crop monitoring; Livestock monitoring; Deep learning

农业是国民经济的基础产业，保障着粮食安全和国计民生[1]。（剩余26711字）