Speaker
Description
With the rise of online shopping and e-commerce platforms, the volume of packages passing through EU shipping centres has more than doubled, requiring new techniques and technologies to help prevent smuggling. CosmoPort is a Horizon-EU funded project aimed to deploy muon scattering tomography in EU postal centres, targeting illicit goods such as drugs, tobacco, and firearms. Using a prototype scanner system developed by GScan OÜ, a material classification exercise was conducted to assess the feasibility of discriminating different low-Z materials, based on machine learning. Using training data consisting of 30 minute measurements of 6cm side-length cubes of various low-Z materials, a classification accuracy of greater than 98% was achieved with this prototype system. This presentation will discuss the results of this exercise, the application of machine learning techniques to the low-Z problem, and ongoing developments in reliable material identification.