Speaker
Description
DeepAngle uses machine learning to determine contact angles between different phases in the tomography images of porous materials. The measurement of these angles in 3D can be inaccurate and time-consuming due to the discretized space of image voxels. A computationally intensive solution involves fitting and vectorizing all surfaces using an adaptable grid to measure angles between the desired vectorized planes. However, the present study offers an alternative low-cost technique that utilizes deep learning to estimate interfacial angles directly from images. DeepAngle was tested on synthetic and realistic images and was found to improve the r-squared of predicted angles by 5 to 16%, while reducing computational costs by 20 times. This rapid method is particularly useful for processing large tomography data and time-resolved images that are computationally intensive. The developed code and the dataset are available in a public repository on GitHub at [https://www.github.com/ArashRabbani/DeepAngle].
Note: An extended version of this poster has been accepted for publication by in Journal of Geoenergy Science and Engineering.
| Participation | In-Person |
|---|---|
| Country | United Kingdom |
| MDPI Energies Student Poster Award | No, do not submit my presenation for the student posters award. |
| Acceptance of the Terms & Conditions | Click here to agree |
