Foxtail millet is a vital grain whose amino acid content affects nutritional quality. Traditional detection methods are destructive, time-consuming, and inefficient. This work established a rapid and non-destructive method for detecting essential amino acids in the foxtail millet. To address these limitations, this study developed a rapid, non-destructive approach for quantifying eight essential amino acids—lysine, phenylalanine, methionine, threonine, isoleucine, leucine, valine, and histidine—in foxtail millet (variety: Changnong No. 47) using near-infrared hyperspectral imaging. A total of 217 samples were collected and used for model development. The spectral data were preprocessed using Savitzky–Golay, adaptive iteratively reweighted penalized least squares, and standard normal variate. The key wavelengths were extracted using the competitive adaptive reweighted sampling algorithm, and four regression models—Partial Least Squares Regression (PLSR), Support Vector Regression (SVR), Convolutional Neural Network (CNN), and Bidirectional Long Short-Term Memory (BiLSTM)—were constructed. The results showed that the key wavelengths selected by CARS account for only 2.03–4.73% of the full spectrum. BiLSTM was most suitable for modeling lysine (R² = 0.5862, RMSE = 0.0081, RPD = 1.6417). CNN demonstrated the best performance for phenylalanine, methionine, isoleucine, and leucine. SVR was most effective for predicting threonine (R² = 0.8037, RMSE = 0.0090, RPD = 2.2570), valine, and histidine. This study offers an effective novel approach for intelligent quality assessment of grains.

Foxtail millet is a nutritionally important cereal whose fat and protein content directly influence its nutritional quality and processing properties. To overcome the limitations of traditional detection methods, developing rapid, non-destructive, and interpretable models is essential. A total of 214 samples of the foxtail millet cultivar “Changnong No. 47” were used in this study. The Sparrow Search Algorithm was introduced to screen stable key wavelengths by statistically analyzing their selection frequency. Based on the selected wavelengths, quantitative models were constructed using Partial Least Squares Regression (PLS), Random Forest (RF), and Support Vector Machine. The SHapley Additive exPlanations method was employed to quantify the direction and magnitude of contributions of the key wavelengths within the model. Results show the selection of 13 key wavelengths for fat and 15 for protein. The RF model delivered the best prediction for fat content (R_P² = 0.797, RMSE_P = 0.218%, RPD_P = 2.219), while the PLS model performed best for protein content (R_P² = 0.695, RMSE_P = 0.268%, RPD_P = 1.811). The methodology established in this study can not only be applied to the rapid quality assessment of millet but also be extended to analyze the nutritional components of other grains.