Welcome to SPG India

14th Biennial International Conference SPG 2023

Automatic Fault Detection using Semantic Segmentation based UNET Model with a Strong Backbone Network

Published in GEOHORIZONS - 2023

  • Vol. Vol.29 No.2, Page 1
  • ISSN NO :
  • DOI Link
Rahul Mahadik, Aurobinda Routray, Gagandeep Singh (IIT Kharagpur), Sanjai Kumar Singh

Abstract


Seismic faults are among the core geological features of interest for the geophysicist for efficient seismic interpretation. They are formed by the displacement of the adjacent blocks of rocks. Because faults may seal the porous reservoir rocks, they indicate the probable availability of hydrocarbons. Accurate fault detection is the utmost important step as drilling well is a very expensive venture, and false fault detection may result in drilling dry well and cause huge losses. The seismic survey area is also expanding to fulfill the increasing demand for petroleum products with the growing population. The traditional methods require interpreters and geophysicists to label each fault manually, which is practically tedious, cumbersome, and time-consuming. Moreover, human intervention is prone to errors. Hence, it is paramount to reduce such possible howlers while detecting seismic faults.

Keywords


Deep learning, DENSENET121, Fault accuracy index, Fault interpretation, INCEPTION-v3

Full Article

View Document