Seeds are the foundation of agriculture and are crucial for sustaining species.The selection of high viability seeds through appropriate methods plays a crucial role in agricultural production.Tradi-tional seed viability detection methods such as germination experiments and TTC staining methods were destructive and time-consuming,and couldnot meet the needs of modern agriculture.In recent years,with the deepening of research on detection technology,various non-destructive testing meth-ods have been developed and introduced into seed viability detection.These technologies have advanta-ges such as non-destructive and high-throughput.This article reviewed the molecular vibration effects of substances contained in seeds and the volatilization of viability metabolism during storage from both optical and electrochemical perspectives,including near-infrared spectroscopy,multispectral,hyper-spectral,Raman spectroscopy,and the application of electrochemical techniques represented by elec-tronic noses in seed viability detection.The focus was on describing the content of these technical methods in different research objects,technology research and development,model algorithm devel-opment,and other aspects.In conclusion,this paper analysed the development prospects of artificial intelligence in the field of seed viability detection.It proposed integrating hyperspectral detection tech-nology and electrochemical detection technology to construct a dataset that fused multimodal informa-tion.Coupled with AI algorithms,this approach represented a new direction for seed viability detec-tion,aiming to achieve more accurate results in predicting seed viability for both single crop species and across species.
关键词
作物种子/活力/无损检测/多模态信息融合/人工智能
Key words
crop seed/viability/non-destructive testing/multimodal information fusion/artificial intelligence
维普
万方数据