Data from: Geo-SegNet: A contrastive learning enhanced U-Net for geomaterial segmentation

The dataset was obtained using high-resolution X-ray micro-CT scans with a TESCAN UniTOM XL scanner at Duke University’s Creativ Engineering Laboratory. Samples were secured with a custom core holder featuring a magnetic aluminum base and stabilized using an acrylic tube. Scanning parameters included 160 kV voltage, 20 W power, 4096 projections, 20 µm voxel size, and a 1.5 mm copper filter. The UniTOM XL’s helical scanning captured full core widths, and 9266 representative slices were selected to optimize computation and train models on pore feature segmentation.