With the rising popularity of whey protein as a dietary supplement, ensuring its quality has become imperative for consumer protection. Unscrupulous merchants sometimes adulterate whey protein with inexpensive vegetable protein to boost profits. Despite the criticality of this concern, reliable studies and relevant practical detection methods are currently limited. To fill this gap, this study adopted an integrated technique combining mid-infrared spectroscopy with machine learning to rapidly and accurately identify both single and multiple vegetable protein adulterants in whey protein. First, various recognition models were trained using AdaBoost-support vector classification (AdaBoost-SVC), AdaBoost-decision tree, K-nearest neighbor, SVC, and Gaussian Naive Bayes. Ten-fold cross-validation was subsequently used to determine the optimal spectra pre-processing combination, which included standard normal variate, first derivative, and Savitzky– Golay smoothing. Feature selection was then performed using the successive projection algorithm, principal component analysis, genetic algorithm (GA), and interval partial least squares with GA (iPLS-GA). The classification results revealed that the iPLS-GA-AdaBoost-SVC achieved the best performance on both the training and prediction sets, demonstrating the ability of the iPLS-GA to improve model stability and robustness. Overall, our findings underscore the potential applicability of the proposed method as an accurate and practical tool for improving the quality control of whey protein.