Precision seed certification through machine learning

Abstract

Original and pure seeds are the most important factors for sustainable agricultural production, development, and food security. Conventionally, seed protection and certification programs are carried out on several classes from breeders to certify seed based on a physical, biochemical, and genetic evaluation to approve seed as a cultivar. In seed industries, quality assurance programs depend on different methods for certifying seed quality characteristics such as seed viability and varietal purity. Those methods are mostly conducted in a less cost-effective and timely manner. Combining machine learning (ML) algorithms and optical sensors can provide reliable, accurate, non-destructive, and quick pipelines for seed quality assessments. ML employs various classifiers to authenticate and recognize varieties through K-means, Support Vector Machines (SVM), Discriminant Analysis (DA), Naive Bayes (NB), Random Forest (RF) and Artificial Neural Networks (ANNs). In recent years, progress in ANN algorithms as deep learning simplifies big data analytics procedures by categorizing learning and extracting distinct levels of multiplex data. Deep learning opened a new door for developing a smartphone as a fast and robust substitute for the online seed variety discrimination stage through developing a Convolutional neural networks (CNNs) model-based mobile app. This review presents machine learning and seed quality assessment areas to recognize and classify seeds through long-standing and novel ML algorithms.

Keywords

• high-throughput phenotyping
• machine learning
• seeds certification
• ml algorithms
• artificial neural networks (anns)
• convolutional neural networks (cnns).