Artificial intelligence-based tools for next-generation seed quality analysis

Abstract

Innovation in agrotechnologies is urgently needed to fulfill the demand burden on food and agriculture industries. The key challenge in producing a high-quality, high-yielding crop is using quality seed and its identification. Seed quality identification in the seed industry often uses traditional methods based on manual observations, which are cumbersome and time-consuming. Still, there is always the risk of faulty reporting and non-uniformity in test results among different testing agencies. Because of the changing requirements of the seed industry, Artificial Intelligence (AI)-based tools and various methods have been developed to test the quality of seeds. AI-based tools have been extensively applied in different farming applications. This review explores these tools and strategies, including traditional, semi-automatic, or automated ones developed using machine learning. These include non-destructive techniques such as x-ray imaging, remote sensing, multispectral imaging, hyperspectral imaging, and near-infrared (NIR) spectroscopy, which are less expensive and time and/or labor-savings. Furthermore, we discuss the characteristics of AI-based techniques for depth analysis and their application in various aspects of seed quality, including seed vigor, seed health, seed germination, and seed viability. Lastly, we furhter evaluate the challenges of these methods and how they will provide healthy seeds to each farmer in the future and increase the overall production of crops. We propose to leverage AI-based tools to bridge the knowledge gap between traditional screening methods and integration of advanced technologies for better screening of crop seeds.