In a recent study, laser-induced breakdown spectroscopy (LIBS) combined with deep learning techniques was employed to swiftly identify different varieties of soybean seeds. Soybean seed purity is crucial in agriculture and food processing, necessitating the need for rapid variety identification. Soybeans are rich in vegetable protein and oil, with different varieties exhibiting varying genetic purity and qualities. Mixed or adulterated soybean seeds can pose challenges to farmers and food processors. Various genetic methods like polymerase chain reaction (PCR) and high-performance liquid chromatography (HPLC) are traditionally used for specific identification of soybean varieties but may involve chemical agents. LIBS technology, on the other hand, is non-invasive and does not require chemical agents, making it a promising alternative for rapid genotype discrimination.
The study proposed a novel approach of pressing soybean seeds into a culture plate filled with rubber sand for LIBS analysis, eliminating the need for time-consuming sample preparation. By collecting three LIBS spectra for each soybean seed and applying machine learning and deep learning algorithms, the researchers achieved high identification accuracies. Support vector machine (SVM) and convolutional neural networks (CNN) were utilized for classification. CNN, a form of deep learning, showed superior performance in spectral data processing compared to traditional machine learning algorithms like SVM. The study designed new ResNet-based structures, including PCA-ResNet and PCSA-ResNet, which utilized spectral matrices as input data and achieved optimal classification accuracies of 91.75% in the prediction set.
The results demonstrated the efficacy of deep learning coupled with LIBS for rapid identification of soybean seed varieties. The study highlighted the importance of leveraging advanced technologies for agricultural applications, paving the way for on-site and efficient variety discrimination in soybean seeds. The combination of LIBS and deep learning holds great potential for enhancing agricultural product quality assessment and food processing efficiency.
📰 Related Articles
- NSW Government Initiative Enhances Vocabulary Skills Through Blended Learning
- Deep Learning Fusion Model Enhances Breast Lesion Diagnosis
- Deep Learning Enhances Thyroid Nodule Diagnosis Accuracy with Motion Correction
- Deep Learning Enhances Fetal Weight Estimation Accuracy in Ultrasound
- Viridian Financial Group Enhances Efficiency Through Strategic Alignment