Creating Fault Masks in OpendTect OpendTect’s Volume Builder with the Fault Painter option allows you to take interpreted fault planes, or fault sticks and convert them directly...
Creating Fault Masks in OpendTect
OpendTect’s Volume Builder with the Fault Painter option allows you to take interpreted fault planes, or fault sticks and convert them directly to Fault Masks usable for Machine Learning (ML) training. Fault masks are seismic datasets with value 1 at fault positions and value 0 everywhere else. The training set can now be constructed in the usual way in the “Seismic Image Segmentation” workflow of the Machine Learning plugin.
OpendTect supports the following Machine Learning workflows for predicting Fault Likelihood: 1. Apply the pre-trained ML Fault Predictor that is shipped with the software 2. Create your own training data sets from interpreted faults or fault sticks, train a U-Net (2D or 3D) and apply the trained model 3. Optionally: read the training set generated under 2 into your own Python script create your own model, or update dGB’s U-Net, augment the training set, train the model); put the trained model back in OpendTect so that it can be applied in the ML environment from the User Interface.
Here is a link to a presentation with more detailed information.: https://lnkd.in/duiPBuCT
Send us an email if you would like a demo license: info@dgbes.com
#machinelearning #OpendTect
