Research on the visualization method of lithology intelligent recognition based on deep learning using mine tunnel images

• A TensorFlow-based CNN integrated with PyQt5 achieves 98.8% accuracy in lithology recognition from mine tunnel images. • The preprocessing pipeline—gray scaling, Gaussian blurring, and feature dimensionality reduction—substantially improves rock image classification performance. • The intelligent recognition system can be directly linked to roadway support design, enhancing both reliability and cost-effectiveness in mining engineering. • This approach offers a low-cost, automated alternative to hyperspectral remote sensing for field-based rock lithology identification.