System and method for improved geographical data interpretation

A computer-implemented method for interpreting geophysical data utilising an Artificial Neural Network (ANN), performed by electronic operations executed by a computing device, comprising:

• performing a training processing step on at least one training-data set, comprising the steps of:
• (a) generating a first label-data by segmenting said at least one training-data set into different separate regions of at least a first region, representing a known first region having at least one identified geological feature, and a second region, representing a known second region having at least one unidentified geological feature, and a third region, representing an unknown region, wherein said first, second and third regions are non-overlapping and wherein said third region comprises unlabelled voxels, and generating at least one dynamic score for each voxel of said first label-data, each dynamic score corresponding to a distance value from any one of said at least one identified geological feature;
• (b) generating a first ANN model output for a dynamically adaptable Region of Interest (ROI) of said first label-data, said dynamically adaptable ROI including said first and second region but ignoring or excluding said third region, wherein said at least one dynamic score defines a distance-based loss metric applied during training of said ANN so as to weight a training error according to said distance value;
• (c) generating an updated label-data by selecting at least a first portion of any one of said first, second and third region, and assigning and appending at least said first portion to any one of said first, second and third region;
• (d) generating an updated ANN model output for an updated dynamically adaptable ROI of said updated label-data by retraining said ANN using an updated distance-based loss metric corresponding to updated dynamic scores of said updated label-data;
• (e) repeating steps (c) and (d) until a predetermined condition is met, providing a final ANN model output, wherein the predetermined condition comprises at least one of a predetermined number of reiterations or a quality parameter of said ANN indicative of convergence of said training;
• applying said final ANN model output to a seismic target-data set obtained from a seismic survey of a subterranean volume utilising said ANN, generating a desired output data comprising a voxelwise visual representation of said at least one identified geological feature in said subterranean volume.