Industrial paraffin contamination addition to rice is considered a highly observed food safety issue with adverse effects on human health. Therefore, the ability to identify the level of industrial paraffin contamination in rice has become critical. In this study, the industrial paraffin contamination levels (IPCL) in rice were identified by using hyperspectral imaging (HSI) alongside deep learning classification models. We established a deep learning network, SC-HybridSN, for rapid discrimination of rice IPCL using spatial and spectral features. An ablation experiment was conducted to investigate the roles of each module, and comparisons were made with conventional models (PLS-DA, SVM) under various training conditions. The extracted results clearly showed that the SC-HybridSN model exhibited superior classification performance under conditions involving the single data set, multi-source data set, and small sample training. These results suggest that the proposed model has the capability to swiftly and non-destructively identify rice IPCL. Display omitted ●Demonstration of the feasibility of utilizing HSI and deep learning for the identification of IPCL in rice.●SC-HybridSN was constructed to develop IPCL classification model.●The proposed method has achieved an accuracy rate of 97.38–100 % in various modeling scenarios.