Ensuring that food products are safe, nutritious, and of high quality is a critical priority for U.S. Agriculture and food systems

The Department of Agriculture National Institute of Food and Agriculture awarded $728,000 to The Trustees Of The Stevens Institute Of Technology on February 1, 2026, under the Agriculture and Food Research Initiative (CFDA 10.310) to develop rapid, non-destructive methods for assessing food quality and predicting shelf life using ultra-wideband microwave imaging and deep learning technology.

The project combines data science and microwave imaging to evaluate food products without physical damage, enabling assessment of freshness, internal structure, and early deterioration signs. The work will develop trustworthy deep learning models robust enough to apply across multiple food types, analyze large volumes of imaging data, and create digital twins of food products to predict quality changes over time and guide improved storage and preservation practices. The system is designed to reduce food waste, enhance food safety, and support data-informed decision-making across producers, processors, and retailers.

Performance occurs in Hoboken, New Jersey, with a period of performance through January 31, 2029. The award is a Project Grant, assistance type.

The University of Alabama and the University of Maryland, College Park, serve as sub-recipients. Both institutions received sub-award notices in March and August 2026 respectively, performing work under the DSFAS Partnership: Food Quality Evaluation and Prediction Leveraging Multi-Modal Deep Learning and Digital Twin.

However, current methods for assessing food quality often rely on slow, labor-intensive, and/or destructive testing. These challenges affect producers, processors, retailers, and consumers, with implications for economic efficiency, food security, and environmental sustainability. The goal of this project is to address the need for a rapid, non-destructive approach to food quality assessment and prediction that can support timely decision-making and reduce losses from waste and inefficiencies. In this project, we propose to combine data science and ultra-wideband microwave imaging to enable efficient and effective analysis of large-scale food data. This approach will allow food products to be evaluated without cutting or damaging them, enabling rapid assessment of freshness, internal structure, and early signs of deterioration. We will use multiple image resources and develop trustworthy deep learning technology that is robust enough to be applied to multiple food types. By analyzing large volumes of data and creating digital representations (digital twins) of food products, the system will help predict quality changes over time and provide guidance for improved storage and preservation practices. We expect to develop a scalable, efficient tool that supports data-informed food quality management, reduces food waste, enhances food safety, and strengthens the resilience and sustainability of agricultural and food systems.