Transfer learning

Neural Networks pre-training

Synthetic data generator

Additional training of neural networks based on
real seismic data and interpreter’s

ML algorithms have been applied to the synthetic data, gaining ability to recognize

~1 mln synthetic data samples with various conditions make our approach

Initial Data

Restoring reflections in areas of seismic data losses

Significant increase of Signal-to-noise ratio for further detailed interpretations

Ability to detect thin layers in case of low data resolution

Precise determination of fault zones and real amplitude shifts up to several meters

Automatic correlation of weak and discontinuous reflections

AI

Seismic Data Enhancement

Initial Data

Significant increase of Signal-to-noise ratio for further detailed interpretations

Ability to detect thin layers in case of low data resolution

Precise determination of fault zones and real amplitude shifts up to several meters

Automatic correlation of weak and discontinuous reflections

AI

Restoring reflections in areas of seismic data losses

Automated Fault Determination

Initial Data

Restoring reflections in areas of seismic data losses

Significant increase of Signal-to-noise ratio for further detailed interpretations

Ability to detect thin layers in case of low data resolution

Precise determination of fault zones and real amplitude shifts up to several meters

Automatic correlation of weak and discontinuous reflections

AI

Geobodies Detection

Initial Data

Restoring reflections in areas of seismic data losses

Significant increase of Signal-to-noise ratio for further detailed interpretations

Ability to detect thin layers in case of low data resolution

Precise determination of fault zones and real amplitude shifts up to several meters

Automatic correlation of weak and discontinuous reflections

AI

Seismic Facies Identification