Abstract
This study focuses on developing a robust method for determining contact angles within porous media based on direct geometric determination of local contact angles where we explore two primary approaches: directly measuring the angle at three-phase contact lines and using image recognition/machine learning. Our method involves a regression-based algorithm that additionally assigns a confidence value to each calculated angle, enabling us to assess the reliability of the results. Results are compared with several established approaches in the literature. Through numerical simulations and application to experimental data from both synthetic and real porous media, it is shown that the proposed method is a more reliable and efficient approach for determining contact angles and its distribution within porous media at in-situ conditions.