Counting and Classification of Seed Using Machine Learning Methods

Deep learning, machine learning and image processing techniques have become important tools used in facilitating agricultural work developing solutions to different problems the production phase. In this study, a seed number type detection algorithm was developed using YOLO deep architecture, real-time object employing CNN structure AugeLab Studio sofware. With model average loss factor of 0.417 achieved after 3000 iterations. As result analysis, it has been determined that bean classification accuracy varies between 97% 100%, while chickpea 91% 100%. addition, total 11 beans 10 seeds single with 100% accuracy. The results demonstrated AugeLab, software artificial inteligence based techniques, can be by companies, biotechnology laboratories certification institutions counting seeds. It also variety and/or species separation, separating detecting germinated seeds, or proportioning foreign mixtures processes within shorter time less costs.