Online Archive of University of Virginia Scholarship
FOCUS: Facilitating Observation and Classification Using Specimens; Fighting Yesterday’s Wars: How the Defense Industry Keeps Old Weapons Alive27 views
Author
Ribeiro, Nathaniel, School of Engineering and Applied Science, University of Virginia
Advisors
Norton, Peter, EN-Engineering and Society, University of Virginia
Nguyen, Rich, EN-Comp Science Dept, University of Virginia
Abstract
Even as bureaucracy rationalizes procedures and institutionalizes responsibilities, it can also shelter waste and deter reform.
High school students in Charlottesville who have partnered with the University of Virginia’s School of Education and Human Development to study microscopic specimens collected from their local biomes need a method of automatic image classification.. We present Facilitating Observation and Classification Using Specimens (FOCUS), a zero-shot classification with support for an extensible taxonomy, and compare its performance with a baseline supervised approach using InceptionV3. Our small and large models achieve top-1 accuracies of 13.59% and 24.67% on the iNaturalist 2018 validation set respectively, considerably lower than the top-1 accuracy of the baseline InceptionV3: 60.2%. Furthermore, we establish a new dataset for microscopic image classification: MicroNet10K.
In the Russo-Ukrainian War, outmoded fighting vehicles such as the M2A3 Bradley and M1A1 Abrams are still in frontline use. When Ukraine’s substantial military support comes from countries with state-of-the-art technology, how is this possible? Case studies of Bradley and Abrams operators in the Iraq War and in the Russo-Ukrainian War, an analysis of mothballing, and scrutiny of the iron triangle between defense contractors, Congress, and the Pentagon, indicate that obsolete vehicles endure because of contractors’ political influence.
Ribeiro, Nathaniel. FOCUS: Facilitating Observation and Classification Using Specimens; Fighting Yesterday’s Wars: How the Defense Industry Keeps Old Weapons Alive. University of Virginia, School of Engineering and Applied Science, BS (Bachelor of Science), 2025-12-12, https://doi.org/10.18130/8ys3-ks26.
