Rice Grain Classification and Automated Quality Control Using Deep Learning Approaches

In Bangladesh, rice is a staple food and a significant contributor to the country's economy as well.An accurate rice grain classification and identification are crucial for improving agricultural productivity, ensuring food security and facilitating trade.The quality inspection of rice grain holds significance for both local and export markets ensuring the maintenance of high standards.This study aims to revolutionize rice grain evaluation by employing Convolutional Neural Networks (CNNs) to enhance the classification and identification of various rice classes.The system commences with five pre-trained models ( MobileNet, VGG16, InceptionResNet, In-ceptionV3, Xception) .We obtained accuracy of 95% , 96% , 94% , 94% and 96% using VGG16 , MobileNet , InceptionV3 , InceptionResNet and Xception respectively.Additionally, a majority voting method is employed by the system to assemble the predictions derived from several models.We also used prominent image processing methods to process our images.The primary objective is to present an image processing-based approach that effectively categorizes diverse rice grain classes based on various parameters.To compile our extensive dataset we captured images of various rice grains under diverse lighting conditions.Our research unveils an accuracy of 98% .