Computer Vision for Materials Science

In conjunction with the IEEE/CVF CVPR 2024, Seattle, WA, USA.

The schedule is available; the room for the workshop will be updated in May.

The workshop will be in-person.

There will be a virtual option to attend the workshop. Registered virtual attendees will be provided with Zoom link (by the conference) to attend oral presentations and, possibly, panel, and ask questions. There is no mechanism to support posters on Zoom.

Computer vision and machine learning are emerging as critical tools for materials characterization. In materials science, a variety of microscopy image data (e.g., optical microscopy, electron microscopy, x-ray microscopy) are used to understand material properties. Many of these microscopy techniques make it easy and inexpensive to collect large, complex image data sets that can overwhelm the available time of the subject matter experts required to interpret it. Other microscopy techniques, such as analytical spectroscopic imaging, are much more time consuming, but have the potential to be accelerated through joint collection and modeling with more scalable imaging modalities. In addition, large video and 3D datasets (e.g., produced by new high-speed electron detectors and X-ray computed tomography) are nearly impossible to manually quantify.

The aim of the Computer Vision for Materials Science (CV4MS) workshop is to bring together cross-disciplinary researchers to demonstrate recent advancements in machine learning, computer vision, and materials microscopy, and discuss open problems such as explainability, uncertainty quantification, and representation learning in materials microscopy analysis. There will be a focus on the unique challenges and advantages in materials characterization including limited ground-truth labels, multimodal datasets, and physics-based constraints, which could lead to new computer vision frameworks for the broader scientific imaging community.