April 17, 2025
12:15 PM – 1:15 PM
Speaker:
Ankan Bhattacharyya, Computer Science Department, University of Kentucky
Where:
327 McVey Hall
(Zoom link: https://uky.zoom.us/j/82467171189)
Title:
System for the 2D-3D detection of bourbon barrel features-of- interest
Abstract:
During the bourbon aging process, where bourbon is stored in new, charred oak barrels for months or years, distilleries realize a significant amount of product wastage. Some of this loss arises from expected environmental processes (i.e. evaporation of the bourbon through the porous wood of the barrel), while a considerable portion is due to the unexpected structural failure of the barrel itself. While these events have been a part of the bourbon industry since its inception, it is only recently that distilleries have begun to investigate the properties of the barrel which may contribute to and/or lead to product wastage. This is no easy task. Since each barrel is constructed by hand, barrels may vary in a number of complex ways (e.g. dimensions, number of staves, placement of metal hoops), making it difficult to identify the properties which may correlate with specific outcomes. We present an imaging and machine learning system which combines 2D feature detection models and 3D photogrammetric reconstruction to automate the detection and measurement of barrel features on the bourbon production line, enabling us to capture the raw data needed to begin solving the complex problem of bourbon wastage.