SEAM-AI-GPU-Hackathon

SEAM/NVIDIA Applied Geoscience GPU Hackathon 2021

Authors: Matthew Oke, Anuola Osinaike, Olawale Ibrahim, and Gamaliel Adun

This repository contains the Rocky-AI teams first place Test 2 and third place Test 1 solution to the Parihaka Seismic Facies Identification Challenge in the SEAM/NVIDIA Applied Geoscience GPU Hackathon. The goal was a six-way classification of the Parihaka 3D seismic survey, with emphasis on interface and key facies. We implemented the U-Net architecture of the EfficientNet-B0 pretrained network using Python with Pytorch, and achieved 98% training accuracy and 95% test accuracy on internal validation.