REVIEW OF AI METHODS FOR PREDICTING HYDROCARBON DEPOSITS FROM 3D SEISMIC IMAGES

This work investigates how artificial intelligence techniques can be used to identify potential hydrocarbon accumulations using 3D seismic imagery. It delves into current AI progress for interpreting seismic data, covering methods like convolutional and recurrent neural networks, graph neural networks, and transformers. The paper also addresses significant obstacles, such as problems with the availability and quality of training data, difficulties in applying models to new areas, and the complexities involved in understanding the results. Furthermore, it touches upon techniques for preparing seismic images and extracting relevant features, approaches for dealing with sparse data, and suggests potential avenues for future investigation in this domain.