All Content
Back to top

In-Person Talk: On the Use of Machine Learning for Imaging and Monitoring of Complex Dynamics

Date: Tuesday, 20 June 2023
Time: 2.30pm
Venue: LT15 (North Spine, NS1-04-07), NTU

Machine learning kernels have been employed extensively by now in many areas of vision and computational imaging. In recent work, we have examined a particular application to complex dynamics, e.g. drying processes used in the pharmaceutical industry, where automation and cost are especially desirable. The objective is to characterise how a population of particles evolves and interacts with its environment during the process. Traditionally, machine vision methods first try to identify individual particles (e.g. by sophisticated segmentation methods), classify them and then count them. Instead, we have relied on a statistical approach, where the ensemble properties are collected in the far field, without identifying individual particles. We will review the Peace method (Physics enhanced auto correlation estimator) and some ongoing efforts to extend it.

Professor MIT (Department of Mechanical Engineering) & Singapore-MIT Alliance for Research and Technology (SMART)