ImageXD 2023: Recap
29 March 2023
BY VERONICA WOODLIEF
What is ImageXD?
ADSA, along with UC Berkeley’s Institute for Data Science (BIDS), co-hosted the sixth annual ImageXD on March 16-17 in Irvine, California. So-called XDs, or cross-domain events, bring together scientists, researchers, and theorists across many disciplines. This event included experts in ecological monitoring, sustainable agriculture, biomedical imaging, aerospace engineering, and machine learning, just to name a few.
The common thread among the ~50 participants is that they all work with images as a primary source of data. Researchers who use image processing tools and techniques typically don’t have the opportunity to collaborate with others from different domains working on similar problems. ImageXD was founded in 2016 to address those challenges and provide attendees the opportunity to build collaborations and learn from one another.
This year’s event featured talks and presentations by Sara Beery - Assistant Professor at MIT and Visiting Researcher at Google; Harry Chao - Assistant Professor of Computer Science & Engineering at The Ohio State University; Dan Chitwood - plant biologist and Assistant Professor in the Departments of Horticulture and Computational Mathematics, Science & Engineering at Michigan State University; Kira Evans - Software Engineer at the Chan Zuckerberg Initiative; Catherine Nakalembe - Africa Program Director for NASA Harvest and Assistant Professor at the University of Maryland’s Department of Geographical Sciences; Sharmila Majumdar - UCSF professor and Director of the Center of Intelligent Imaging; and Stéfan van der Walt - Researcher at UC Berkeley and Founder of scikit-image.
What makes ImageXD unique?
Unlike traditional conferences with the overall theme and session topics determined by the organizers, ImageXD is an “unconference”, with informal sessions driven by the attendees’ research interests and goals. The attendees proposed and voted on breakout session topics over coffee the first morning. Some of the resultant projects included “What Even is Geospatial Data?”, “Open Data — How Does it Work?”, “Python Packaging”, and “Decolonizing Data”.
Although it was a tech-focused event, the first day ended with good old-fashioned, hand-drawn poster presentations crafted with sharpies and sticky notes. The mood was enthusiastic - the groups made significant progress on their projects. One attendee reported feeling so inspired and energized by the day’s accomplishments that she compared it to being on vacation - not a common sentiment after a long day at a conference! Team “Open Data” even declared (maybe slightly tongue-in-cheek) that they had solved the pressing issue of sharing and managing large datasets. On the event's second day, most groups continued working on their projects from the previous day rather than starting new ones - a testament to their enthusiasm and the momentum they had already gained.
Outcomes & Impacts
By the end of the second day, attendees had learned new skills, addressed complex challenges, established best practices for data management, and even designed an entirely new software package - all in just two days. The connections that were made - across disciplines, backgrounds, and even career stages - are the explicit purpose of ImageXD. The goal is for those connections to continue beyond the event, with ongoing collaboration that will have a lasting impact on the image processing community.