A Benchmark Dataset for Semi-Automatic Seismic Interpretation Based on a New Zealand's Seismic Survey

In the past decades, machine learning algorithms have been developed for accelerating the seismic interpretation process. For this purpose, benchmark availability for algorithm improvement in computational geophysics is imperative. However, in geosciences, the supply of open benchmark datasets is very limited. The purpose of our study was to evaluate a CNN algorithm response in comparison to an OpendTect tool for seismic horizon interpolation, and in addition, offer a benchmark dataset for further algorithms evaluation.