From Paddy to Pixel: An In-depth Exploration into Classifying Diverse Rice Varieties Leveraging Advanced Convolutional Neural Network Architectures

Rice, one of the most extensively cultivated grain commodities globally, has a diverse range of genetic variations. These different types are distinguished from one another based on specific characteristics they possess. By utilizing the distinguishing characteristics of rice types, it becomes feasible to categorize and assess the standard of seeds. Rice varieties need to be distinguished for several reasons. Like, every type has different requirements for water, nutrients in the soil, and resistance to diseases and pests. Accurate rice variety identification allows farmers to implement specialized agronomic techniques that increase production and reduce resource waste and different varieties of rice satisfy different consumer preferences. For this reason, this research has proposed a Convolutional Neural Network framework to classify different varieties of rice. The dataset used in this study consisted of 75,000 photos, which were obtained from the comprehensive Rice Image Dataset. These images represented five different rice varieties. The initial findings demonstrate a notable level of high accuracy in classification of 98.88%, indicating the practicality and resilience of the methodology implemented. The model has achieved an average precision of 97% and an f-1 score of 0.97. Acquiring such knowledge will enable individuals to make well-informed decisions about cultivation, harvest, post-harvest processing, and market dynamics, thereby enhancing agricultural efficiency.