Swin-UNETR: A transformer-based model for 3D pore network segmentation in low-permeability sedimentary rocks

Key Points

• The new model achieves superior 3D segmentation accuracy for pore networks, enhancing understanding of sedimentary rock structures.
• A key improvement is seen with a segmentation accuracy of 92% compared to traditional methods in low-permeability rocks.
• Assessment using a transformer-based model shows its effectiveness in recovering complex pore structures better than older techniques.
• The findings highlight the model's relevance for geological studies, suggesting further exploration in real-world sedimentary formations.