Objective Core logging is crucial for obtaining deep geological information about the earth. Currently, manual logging remains the primary method for acquiring lithological and other information, but it is time−consuming, labor−intensive, and prone to incomplete results and subjectivity.

Methods Therefore, this study focuses on the ZKH3 of sandstone−type uranium deposits in the southwest of the Ordos Basin, applied deep learning and imaging spectroscopy techniques to core lithology identification. This study constructed a CNN model consisting of 7 one−dimensional convolutional layers, 2 pooling layers, 1 one−dimensional CBAM, and 3 fully connected layers.

Results Additionally, a total of 26877 spectral samples from 7 types of rocks were collected, and model optimization and training were completed. Finally, the performance of the model was evaluated through comparison with Support Vector Machine (SVM) and its application in the whole borehole.The results show that the overall accuracy (OA) of deep learning model reached 94.6%, among which the producer’s accuracy (PA) of mudstone, fine sandstone, siltstone, medium sandstone, coarse sandstone, glutenite and background were 95.07%, 72.02%, 97.50%, 97.37%, 96.65%, 97.33%, and 99.01%, respectively. The Kappa coefficient was 0.94, which was better than SVM overall and achieved comparable results to geological logging.

Conclusions This indicates that deep learning model based on imaging spectral data demonstrates excellent lithology classification and identification capabilities for core samples. And this approach enables non-destructive and rapid lithology identification while reducing the subjectivity inherent in manual geological logging to some extent, which provides valuable reference for digitization and automated logging research of core samples reference for research on digitalization and automation of core logging.