SoyNet: Deep Learning Approaches for Automated Soybean Seed Quality Assessment

In the realm of agriculture, soybeans have played a vital role in diets for centuries, providing a consistent source of plant protein. Beyond protein, soybeans offer essential components like fibres, vitamins (including B vitamins), minerals (such as iron, calcium, and magnesium), and antioxidants. This necessitates precise quality assessment for factors like size, shape, colour, and damage. Leveraging advanced machine learning techniques such as VGG16, AlexNet and CNN, the model aims to automate the identification and categorization of soybean quality. To implement the proposed work the standard soybean dataset is considered. It consists of five soybean seed types complete, spotted, immature, broken, and skin-damaged. The proposed work effectively categories the quality of soybean seeds and it has the potential to improve the agricultural process as compared to traditional work. In this work different pretrained models are experimented in that VGG16 provides the highest of all showing accuracy 90%, AlexNet accuracy 78.49% and CNN accuracy 74.55%.