Published by
SPE (Onepetro.org) - https://doi.org/10.2118/228675-MS
Summary
This study introduces a CNN-based U-Net framework for seismic facies classification, trained on labeled data and tested on an unlabeled dataset. Evaluation metrics confirmed the model’s accuracy and efficiency, demonstrating its ability to reduce manual effort, enhance reliability in subsurface analysis, and enable scalable deployment in oil and gas exploration.