Seeing What's Hard to See: AI for Agricultural Infrastructure Understanding at Scale

Oishee Hoque
UVA CS

Time: 2026-04-08, 12:00 - 13:00 ET
Location: Rice 540 and Zoom

Abstract Effective agricultural and environmental monitoring depends on accurate, up-to-date maps of infrastructure, yet this information is often incomplete, outdated, or difficult to obtain at scale. Satellite imagery offers a promising path toward scalable monitoring, but extracting meaningful information presents significant challenges: relevant structures are sparse and difficult to detect, labeled data is limited, and infrastructure exhibits complex spatial dependencies that standard vision models fail to capture. In this talk, I will introduce the kinds of work we are doing at the Biocomplexity Institute at UVA, focusing on AI for agricultural and environmental monitoring. I will discuss our work on building large-scale datasets and developing methods that go beyond standard deep learning by incorporating domain knowledge and structural constraints. Throughout, I will highlight both the opportunities and the practical challenges of applying AI to environmental monitoring, including issues of generalization, annotation cost, and what it takes to move from research datasets toward practical impact.

Bio: Oishee Bintey Hoque is a Ph.D. candidate in Computer Science at the University of Virginia, specializing in Knowledge-Guided Machine Learning, Multimodal Understanding, and Remote Sensing. Her research focuses on developing large-scale datasets and machine learning frameworks that integrate domain knowledge to improve model interpretability, scalability, and performance across complex real-world systems. She has also interned at the USDA-NIFA/NSF AI Institute for Next Generation Food Systems, where she contributed to geospatial data processing and model benchmarking for benchmarking for agricultural AI. She is a recipient of the UVA School of Engineering and Applied Science Endowed Fellowship, awarded to outstanding doctoral students for research excellence.

Prior to joining UVA, Oishee worked as a Software Engineer at Enosis Solutions in Bangladesh. She received her B.Sc. in Computer Science and Engineering from Ahsanullah University of Science and Technology, Bangladesh.