Understanding the dynamic fracturing and deformation behaviour of geomaterials, such as concrete and rock, is essential for underground infrastructure safety. This study integrates experimental techniques, including the Triaxial Hopkinson Bar (Tri-HB) system, digital image correlation (DIC), digital volume correlation (DVC), acoustic emission (AE), and high-speed X-ray phase contrast imaging (XPCI), to analyse the mechanical and fracturing properties of geomaterials under high strain rates. The results reveal the interplay between stress confinement, strain rates, and microcrack evolution. A machine learning-based crack classification method is proposed to distinguish crack types and their evolution. This study provides a foundation for numerical modelling and further engineering applications.