seismiqb is a framework for deep learning research on 3d-cubes of seismic data. It allows to

• sample and load crops of SEG-Y cubes for training neural networks
• convert SEG-Y cubes to HDF5-format for even faster load
• create_masks of different types from horizon labels for segmenting horizons, facies and other seismic bodies
• build augmentation pipelines using custom augmentations for seismic data as well as rotate, noise and elastic_transform
• segment horizons and interlayers using UNet and Tiramisu
• extend horizons from a couple of seismic ilines in spirit of classic autocorrelation tools but with deep learning
• convert predicted masks into horizons for convenient validation by geophysicists

git clone --recursive https://github.com/gazprom-neft/seismiqb.git

Seismic cube preprocessing: load_cubes, create_masks, scale, cutout_2d, rotate and others.

Solving a task of binary segmentation to detect seismic horizons.

Extending picked horizons on the area of interest given marked horizons on a couple of ilines/xlines.

Performing multiclass segmentation.

Please cite seismicqb in your publications if it helps your research.

Khudorozhkov R., Koryagin A., Tsimfer S., Mylzenova D. Seismiqb library for seismic interpretation with deep learning. 2019.

@misc{seismiqb_2019,
  author       = {R. Khudorozhkov and A. Koryagin and S. Tsimfer and D. Mylzenova},
  title        = {Seismiqb library for seismic interpretation with deep learning},
  year         = 2019
}