FaultSSL: Seismic fault detection via semisupervised learning

The prevailing methodology in data-driven fault detection leverages synthetic data for training neural networks. However, it grapples with challenges when it comes to generalization in surveys exhibiting complex structures. To enhance the generalization of models trained on limited synthetic data sets to a broader range of real-world data, we introduce FaultSSL, a semisupervised fault detection framework. This method is based on the classical mean teacher structure, in which its supervised part uses synthetic data and a few 2D labels. The unsupervised component relies on two meticulously devised proxy tasks, allowing it to incorporate vast, unlabeled field data into the training process. The two proxy tasks are panning consistency (PNC) and patching consistency (PTC). PNC emphasizes feature consistency in overlapping regions between two adjacent views in predicting the model. This allows for the extension of 2D slice labels to the global seismic volume. PTC emphasizes the spatially consistent nature of faults. It ensures that the predictions for the seismic data, whether made on the entire volume or individual patches, exhibit coherence without any noticeable artifacts at the patch boundaries. Although the two proxy tasks serve different objectives, they uniformly contribute to the enhancement of performance. Experiments showcase the exceptional performance of FaultSSL. In surveys wherein other mainstream methods fail to deliver, we present reliable, continuous, and clear detection results. FaultSSL reveals a promising approach for incorporating large volumes of field data into the training and promoting model generalization across broader surveys.