About this Abstract |
Meeting |
TMS Specialty Congress 2024
|
Symposium
|
2nd World Congress on Artificial Intelligence in Materials & Manufacturing (AIM 2024)
|
Presentation Title |
A Materials Data Segmentation Benchmark (MDSB)
|
Author(s) |
Pawan Kumar Tripathi, Kristen Hernandez, Tommy Ciardi, Mingjian Lu, Max Liggett, Andrew Ballen, Jean-Baptiste Forien, Brian Giera, Manyalibo J. Matthews, Mengjie Li, Kristopher Davis, John Lewandowski, Laura Bruckman, Yinghui Wu, Roger French, Vipin Chaudhary |
On-Site Speaker (Planned) |
Vipin Chaudhary |
Abstract Scope |
This work focuses on creating Materials Data Segmentation Benchmark (MDSB), designed to provide diverse materials datasets, ground truth, and metrics for evaluation of models for benchmarking purposes. The MDSB library will serve as a centralized directory/repository of diverse materials datasets, containing various imaging modalities and representing a wide range of materials. It provides rich collection of data encompassing materials from metals and alloys to polymers. The repository ensures that segmentation models can be rigorously evaluated on a diverse range of materials and imaging techniques. MDSB will curate a wide array of materials datasets, including 2D and 3D data, along with various imaging techniques such as microscopy, tomography, and radiography with the annotated ground truth/gold standard. MDSB will be an open-source project, freely accessible to the research community. This will encourage innovation, democratize access to advanced segmentation tools, and accelerate the development of new materials analysis techniques. |
Proceedings Inclusion? |
Definite: Other |