Machine Learning and Sensor Data Fusion Approach for Nondestructive Multivariate Classification of Codling Moth Infested Apples

Grants and Contracts Details


We propose the development of a noninvasive approach for detection and classification of codling moth (CM) infested apples based on machine learning-big data analysis of fused sensor data from hyperspectral imaging (HSI) and Acoustic Emission (AE) systems. CM infestation is the most devastating pest problem confronting the US apple industry, despite pesticide control and other mitigation efforts. One CM infested apple found in a shipment can reduce return to producers by 59%. In the US, incidents of CM detected in apple truckloads increased by 276% in recent years. There are currently no methods to rapidly scan every apple noninvasively, leaving producers vulnerable to rejection or devaluation of their shipments. The main goal of this study is to develop rapid and noninvasive approach to detect and classify CM infestation in post-harvest apples, by using pattern recognition algorithm on features from fused HSI and AE scans. This would allow objective assessment of every apple for infestation. Our specific objectives are to: 1. Authenticate and characterize AE signal source from CM infested apples. 2. Determine the impact of apple storage conditions on CM acoustic signal strength 3. Identify multivariate analytical approach based on machine learning and fusion of AE and HSI sensor data to detect and classify CM infested apples. Our multidisciplinary team has the expertise to address the above objectives and help the multibillion dollar US apple industry remain globally competitive and sustainable. With this, we will fulfill one of the aims of USDA-NIFA Foundational Program under Agricultural Engineering priority area (A1521).
Effective start/end date6/1/195/31/23


  • National Institute of Food and Agriculture: $473,989.00


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.