<p>Maturity assessment is a key link in the precise grading and sorting of edamames with pod. Traditional maturity assessment of edamames with pod relies on manual visual inspection by agricultural producers, which suffers from high labor intensity, strong subjectivity, and high error rates. This study employed visible-near infrared hyperspectral imaging technology to perform non-destructive, precise detection on edamames with pod samples representing four maturity stages. In this study, linear partial least-squares discriminant analysis (PLS-DA), least-squares support vector machine (LS-SVM), and k-nearest neighbors (KNN) were applied to discriminate the maturity of edamames with pod. In addition, EMSTFormer-SpecNet of maturity assessment model for edamames with pod was proposed in this study, which included a channel attention mechanism, a multi-scale feature extraction module, and a transformer module. The result of this study demonstrates that the EMSTFormer-SpecNet model outperforms machine learning models and common deep learning models in processing one-dimensional hyperspectral reflectance data of edamames in pod with different maturity levels. The classification accuracy rate of the test set reached 97.82%, which was 9.93% higher than that of the optimal machine learning model SNV-CARS-PLS-DA, and 7% and 4.38% higher than that of the deep learning models 1DCNN and 1D-ResNet, respectively. The result of this study demonstrates that hyperspectral imaging technology combined with deep learning methods has great potential in the maturity assessment of edamames with pod, providing a theoretical basis and technical support for the accurate grading and sorting of podded food.</p>