Abstract:Fault identification plays a vital role in geological structure interpretation, reservoir characterization, and reservoir evaluation. The traditional process for fault identification is label intensive. Deep learning-based, automatic fault detection from seismic data has attracted extensive attention in recent years. However, using a 2D neural network (i.e., single seismic data slice) leads to the loss of correlation between seismic slices, and using a 3D neural network (i.e., 3D seismic data cube) increases the workload of fault labeling and the difficulty of training. To overcome the drawbacks of 2D and 3D neural networks, in this work, we introduce a channel attention mechanism in the original U-net architecture. The new model, which is referred to as 2.5D channel attention U-net (2.5D CAU-net), takes four adjacent seismic slices as inputs instead of the entire 3D seismic data cube. The efficacy of 2.5D CAU-Net for fault detection is demonstrated using publicly available synthetic seismic datasets and real seismic field datasets. We explored the influence of cropping the original seismic data into various sizes on fault identification. Experimental results show that the 2.5D CAU-net can effectively utilize the correlation information among adjacent seismic slices to improve fault detection performance. For different seismic data cropping strategies, results indicate that the more extensive size cropping method allows the network to obtain more contextual information and more complete fault features, improving the efficacy of fault picking. However, the cropping approach with a larger size leads to a smaller number of labeled examples. When the amount of training data is not large enough, the larger size cropping approach tends to lead to model over-fitting, which will decrease the performance of the fault recognition, especially in a 3D neural network. Overall, 2.5D CAU-net can improve the efficiency of fault recognition with less annotation effort and computational resources.

3D Fault Detection: Using Human Reasoning to Improve Performance of Convolutional Neural Networks

Seismic Fault Detection with Convolutional Neural Network

Deep Convolutional Neural Network for Automatic Fault Recognition from 3D Seismic Datasets

Data augmentation for 3D seismic fault interpretation using deep learning

3-D Seismic Fault Detection Using Recurrent Convolutional Neural Networks With Compound Loss

CNN Based Fault Recognition with Multi-Scale Fusion Attention Mechanism

Fault Detection via 2.5D Transformer U-Net with Seismic Data Pre-Processing.

Fault Recognition Method Based on Attention Mechanism and the 3D-UNet

Seismic fault detection using convolutional neural networks with focal loss

3D seismic Fault Detection via Contrastive-Reconstruction Representation Learning

A method for intelligent identification of faults in seismic using an attention-based ES-UNet network with model re-training learning

Nonlocal Structure Constrained Neighborhood Preserving Embedding Model and Its Application for Fault Detection

Fault Detection by Using Instance Segmentation

Interpretability-Guided Convolutional Neural Networks for Seismic Fault Segmentation

Automatic Geological Fault Identification from Seismic Data Using 2.5D Channel Attention U-net

Visual Attention-Guided Learning With Incomplete Labels for Seismic Fault Interpretation

Improving Performance of Seismic Fault Detection by Fine-Tuning the Convolutional Neural Network Pre-Trained with Synthetic Samples

Fault3DNnet: A Lightweight 3D Seismic Fault Detection Network with Bidirectional Decoding

Seismic Fault Interpretation Using 3-D Scattering Wavelet Transform CNN

Automatic Geologic Fault Identification from Seismic Data Using 2.5D Channel Attention U-net

MTL-FaultNet: Seismic Data Reconstruction Assisted Multitask Deep Learning 3-D Fault Interpretation.