With over 30 years of experience in Machine Learning, dGB Earth Sciences continues to innovate in the field of seismic reservoir characterization.
Today, we take a trip down memory lane to showcase our journey and highlight our latest breakthroughs.
Back in the early nineteen-nineties, our co-founders Paul de Groot and Bert Bril were instrumental in a groundbreaking project at TNO, The Netherlands' foremost R&D organization. This project involved the use of supervised and unsupervised neural networks and stochastic pseudo-well modeling for seismic reservoir characterization. The immense success of this project paved the way for the establishment of dGB Earth Sciences in 1995.
In 1998, Equinor (formerly Statoil) sought our expertise in tackling a geohazard interpretation problem. Collaboratively, we developed the ChimneyCube, a cutting-edge supervised neural network-based technique. This solution empowered Equinor to effectively interpret fluid migration paths, assess trap integrity, and identify potential leakage. The positive outcome of this collaboration led to Equinor funding the development of d-Tect, a seismic pattern recognition and attribute processing system.
Fast forward to 2003, we embraced the open-source philosophy by making d-Tect available as OpendTect + plugins. Among these plugins, our Neural Networks plugin gained widespread acclaim for its exceptional performance in both supervised and unsupervised workflows. Building upon this foundation, we are thrilled to unveil our new Machine Learning (ML) plugin in 2020, equipped with the latest deep learning algorithms.
Our ML plugin encompasses all the capabilities of the original Neural Networks plugin while introducing exciting new features for seismic, wells, and wells-to-seismic applications. We pride ourselves on delivering a comprehensive solution that caters to the diverse needs of geoscientists at every level. Here are some unique selling points of our ML plugin:
- Only System with User Interface Driven workflows and Programming Options
- Easy workflows for Operational Geoscientists
- Flexible labeling options for Experimental Geoscientists
- Python Environment for R&D Geoscientists
- SynthRock stochastic simulator plugin to create examples for training
- Growing library of powerful pre-trained models
- Possibility to share proprietary pre-trained models within a company