Sara McMains


In this talk, I will review some of our recent results formulating computer vision problems in manufacturing and microscopic image analysis to exploit various computational geometry concepts and tools. I will show how we apply the concept of alpha-hulls to formalize a mathematical definition of the boundaries of under-reinforced areas in cross-sectional images of composites, enabling quantitative statistical summaries. I will present our specialized Voronoi diagram construction algorithm for detecting such defects in 3D-printed fiber-reinforced polymers. I will show how we use watershed segmentation and the circular Hough Transform to automatically characterize spherical metal powders for additive manufacturing. I will present an algorithm that builds on topological skeletons to automate microstructure grain size characterization of metallic materials, following international standards designed for manual image inspection. Finally, I will introduce the contour gradient charts we developed for recognizing the breakage of carbon fibers in image cross-sections. We have implemented these techniques to successfully analyze hundreds of microscope images from different industries, with all their artifacts and ambiguities, meeting the challenges that real-world data entails.


Sara McMains is a Professor in the Department of Mechanical Engineering, University of California, Berkeley. Her research interests include Geometric DFM (Design for Manufacturing) feedback, computational geometry, geometric and solid modeling, GPU algorithms, 3D printing, CAD/CAM, Geometric Dimensioning and Tolerancing (GD&T), and deep learning. She received her A.B. from Harvard University in Computer Science, and her M.S. and Ph.D. from UC Berkeley in Computer Science with a minor in Mechanical Engineering. She is the recipient of Best Paper Awards from Usenix (1995) , ASME DETC (2000), NAMRC (2015), and ASEE (2019), a Best Poster and a Best Paper Award from the ACM Solid and Physical Modeling Symposium (2007, 2008 -- 2nd place), the Audi Production Award (2011), and the NSF CAREER Award (2005).