AIseed Simulation: A seed simulation sorting software for rapidly determining seed processing procedures and parameters.

• A software for the establishment of the seed processing process was designed. • It can be used to achieve real-time seed simulation sorting. • It integrates "Simulated sorting" and "One-click sorting" functions. • Verified the performance of the software on five crop seeds. Seed processing is a crucial measure for improving seed quality after harvest. However, traditional seed processing procedures and parameter settings have largely depended on technicians' operational experience. To enhance the digitization of the seed processing procedures, AIseed Simulation software was developed and incorporates ''Simulated sorting'' and ''One-click sorting'' based on RGB images of seeds and impurities. Based on the sorting principles of current main seed sorting equipment, length, width, thickness, specific gravity, and color features (R, G, B, R/G, R/B, G/B) are selected. Through Spearman correlation analysis and probability density distribution plots, the sorting features and parameters are determined. By repeatedly comparing precision, recall, and F1 values of different sorting combinations, simulation sorting on a computer is achieved. The ''One-click sorting'' function has been integrated into the above procedures, facilitating automation and making the determination of optimal seed processing procedures and parameters easier and more rapid. The AIseed Simulation was individually validated on five crop seeds— Astragalus membranaceus , Perilla frutescens , Scutellaria baicalensis , Bupleurum chinense , and Platycodon grandiflorus —and the processing procedures and parameters established for each seed type were validated with actual processing equipment. The results validated that AIseed Simulation is a stand-alone, automated RGB image processing platform suitable for seed processing and parameter setting. Furthermore, the AIseed Simulation software exhibits versatility in a range of seed processing scenarios, suitable for both large-scale seed processing facilities and the demands of scientific research and education. These findings contribute significantly to enhancing seed quality and facilitate the intelligent development of the seed processing industry. [ABSTRACT FROM AUTHOR]