In practical applications, effectively segmenting cracks in large-scale
computed tomography (CT) images holds significant importance for understanding
the structural integrity of materials. However, classical methods and Machine
Learning algorithms often incur high computational costs