Abstract: Specialty milks (camel, donkey, and horse milk) have high nutritional value; however, they encounter issue with adulteration by budget-friendly cow milk and mislabeling their proportions. In this study, near-infrared (NIR) spectroscopy was employed to both discriminate between raw cow milk and specialty milks,as well as quantify adulteration levels in specialty milk-cow milk mixtures. Multiple machine learning algorithms were applied to authenticate four types of milk. The PLS-DA, SVM, KNN, and CNN models achieved perfect classification accuracy (100% ± 0%), while the DT model showed slightly lower performance (98.57% ± 0.78%). Meanwhile, The Grad-CAM visualization revealed that the CNN classifiers distinguished the four milk categories by identifying distinct activation patterns correlated with characteristic absorption bands. For adulteration analysis, both the CNN and PLSR models performed excellent performance in predicting specialty milk adulteration levels with cow milk across a concentration range of 2–80% (v/v), achieving R² values >0.99 and RMSE <1.72%. These findings suggest that NIR spectroscopy combined with CNN machine learning provides an efficient solution for addressing classification and adulteration issues in raw specialty milks , then achieve quality control it in market.