USoil: Development of USCS-Based Soil Classifier Using Digital Image Processing and Convolutional Neural Network

Abstract

The purpose of this study is to demonstrate how image processing and convolutional neural networks may be utilized to efficiently classify a soil sample using the Unified Soil Classification System (USCS). The system is composed of five sections: jar testing, picture capture, image processing, Convolutional Neural Network (CNN) training system, and outcome. The convolutional neural network is a type of machine learning that enables quicker image processing while still providing accurate evaluation and output. The system will be based on picture data collected from soil samples using the camera included in the hardware component. As the default neural network tool through Google Colab, the breakdown will be 70% for training, 15% for testing, and another 15% for validation. The application will then display the projected amount and proportion of each soil type – silt, sand, gravel, and clay – in a given sample, before categorizing it according to the conditions specified for each classification. In all, this study classified soils and was found to be 92.4% accurate.