Related Experiment Video
Updated: Sep 26, 2025
11:37
RGB and Spectral Root Imaging for Plant Phenotyping and Physiological Research: Experimental Setup and Imaging Protocols
Published on: August 8, 2017
16.4K
Identification of Moldy Peanuts under Different Varieties and Moisture Content Using Hyperspectral Imaging and Data
Ziwei Liu1, Jinbao Jiang1, Mengquan Li1
1College of Geosciences and Surveying Engineering, China University of Mining and Technology (Beijing), Beijing 100083, China.
Foods (Basel, Switzerland)
|April 23, 2022
Summary
A new data augmentation method, difference of spectral mean (DSM), improves moldy peanut identification using hyperspectral imaging. This technique enhances model accuracy by reducing moisture content
Area of Science:
- Agricultural science
- Food safety
- Spectroscopy
Background:
- Aflatoxins in moldy peanuts pose serious health risks, necessitating effective screening methods.
- Hyperspectral imaging offers potential for identifying moldy peanuts, but moisture content variations complicate spectral and texture analysis.
- Existing identification models struggle with accuracy due to moisture-induced data variability.
Purpose of the Study:
- To develop and validate a data augmentation method to mitigate the impact of moisture content on moldy peanut identification.
- To enhance the generalization ability and robustness of machine learning models for moldy peanut detection.
- To provide a reference for screening other mold-contaminated food products.
Main Methods:
- Collected near-infrared hyperspectral images of 39,119 peanut kernels across 5 varieties, 4 mold classes, and 3 moisture gradients.
- Developed a data augmentation technique named difference of spectral mean (DSM).
- Evaluated the DSM method using K-nearest neighbors (KNN), support vector machines (SVM), and MobileViT-xs models on datasets with two and three moisture gradients.
Main Results:
- The DSM data augmentation method significantly reduced the negative impact of moisture content differences on identification accuracy.
- DSM demonstrated the highest accuracy improvement across all 5 peanut varieties.
- Model accuracy improvements were observed: KNN by 3.55%, SVM by 4.42%, and MobileViT-xs by 5.9% when using two moisture gradients.
Conclusions:
- The proposed DSM data augmentation method effectively enhances the accuracy and robustness of hyperspectral imaging-based moldy peanut identification.
- This approach offers a valuable solution for improving food safety screening, applicable to peanuts and other commodities like corn, oranges, and mangoes.
