From the perspective of soil engineering, soil is uncertain and heterogeneous. Therefore, if an attempt is made to determine the soil classification of a soil without a precise test, for example, an engineer's individual judgement is often involved in making the determination based on his/her own experiences. In relation to acquiring vast and varied knowledge which is easily influenced by individual experiences, the purpose of this study is to gather the know-how of engineers and to create a certain index for use in making on-site judgments that are likely to be more inclusive of various data than those of individual engineers. This study discusses the potential of image recognition by artificial intelligence, using a machine learning technique called deep learning, for the purpose of expanding the cases which employ artificial intelligence. Deep learning was performed with a model using a neural network in this study. For three types of soil, namely, clay, sand, and gravel, an AI model was created that was conscious of the practical simplicity of the images used. It was shown that artificial intelligence, along with deep learning, can be applied to soil classification determination by performing simple deep learning with a model using a neural network.
ASJC Scopus subject areas