Proposed Approach for Creating Soybean Grain Image Dataset

The integration of digital technologies and artificial intelligence in agriculture has the potential to significantly improve the accuracy and efficiency of grain classification. This study focuses on the development of a comprehensive methodology for soybean grain classification, utilizing a custom-built image acquisition system and advanced image processing techniques. High-resolution images of soybean grains were captured using a Nikon D3100 DSLR camera, with the setup optimized to ensure consistent lighting and contrast for precise image analysis. Various segmentation methods, including RGB and CMYK color channel separation, Otsu thresholding, and edge detection using the Canny algorithm, were employed to isolate and classify key features of the grains. Classical image processing techniques were used to create a robust and labeled dataset, providing essential training data for machine learning models. The results demonstrate the potential of combining classical image segmentation with machine learning to automate grain classification processes, enhancing reliability and ensuring compliance with industry standards.