International Safe Transit Association
Public Video

An implementation of Machine Learning for Packaging Evaluation

Euihark Lee
Assistant Professor
Michigan State University School of Packaging

The packaging evaluation is one of the important processes to ensure the functionality of packages. When a new package is produced, the new package goes through the evaluation process to certify the package performance. Many different evaluation methods have been developed over the last few decades, and package evaluation using mechanical tests have been used widely. The goal of mechanical testing is to simulate the environment that the package will face during distribution and consumer usage. However, mechanical testing is a very costly and time-consuming process.

To resolve these issues, we are introducing several packaging evaluation methods using machine learning algorithms. Machine learning is a subset of artificial intelligence (AI) and it can learn and improve performance automatically without being explicitly programmed. In this presentation, two different machine learning applications will be introduced for the packaging evaluation purpose. The first application is estimating box compression strength using a machine learning algorithm, especially an artificial neural network. The next application is evaluating package failure using consumer reviews in the e-commerce platform. The proof of concept through the proposed methods will be discussed with case studies.