Rice Grain Classification Using Machine Vision Transfer Learning

The study provides an innovative method for rice grain categorization based on Mask RCNN architecture and deep transfer learning methods. To support the agricultural and food processing sectors, the study must tackle the difficulties related to precise and effective rice grain categorization. The suggested model shows impressive performance in correctly identifying and categorizing rice grains across several classes by using a dataset that includes images of five different rice grain varieties: Arborio, Basmati, Ipsala, Jasmine, and Karacadag. With a mean Average Precision (mAP) of 99.5%, the model continuously obtains high detection accuracies during extended assessment. The study results demonstrate the created approach's effectiveness and dependability, underscoring its potential for real-world implementations in agricultural automation, quality assurance, and product differentiation.