Reproducible material for A hybrid approach to seismic deblending: when physics meets self-supervision - Luiken N., Ravasi M., Birnie, C. submitted to NeurIPS 2022.

We are grateful to the authors of the PyTorch implementation of the High-Quality Self-Supervised Deep Image Denoising 2019 NIPS paper, who we have adapted to handle structured noise. Some of the codes provided in ssinterp.model are cleaned-up versions of the original codes.

Project structure

This repository is organized as follows:

• 📂 ssinterp: python library containing routines for deblending by inversion with a self-supervised denoiser;
• 📂 data: folder where the MobilAVO dataset must be placed (note that the data can be downloaded from this link (follow instructions under External links tab))
• 📂 notebooks: set of jupyter notebooks reproducing the experiments in the paper (see below for more details);
• 📂 scripts: set of python scripts used to run multiple experiments with different input parameters for the ablation studies
• 📂 figures: folder where figures from various script experiments will be saved.
• 📂 results: folder where results from various script experiments will be saved for later analysis.

Notebooks

The following notebooks are provided:

• 📙 SSNetwork_impulseresponse.ipynb: notebook displaying the impulse response of the blind-network used in this work;
• 📙 Deblending_CCG-fourier.ipynb: notebook performing benchmark deblending by inversion in CCG domain using a patched fourier sparsity transform;
• 📙 SSDeblending_CCG-denoising.ipynb: notebook performing benchmark deblending by denoising in CCG domain using the proposed self-supervised network;
• 📙 SSDeblending_CCG-pnp.ipynb: notebook performing benchmark deblending by inversion in CCG domain using the proposed PnP algorithm;
• 📙 SSDeblending_CRG-pnp.ipynb: notebook performing benchmark deblending by inversion in CRG domain using the proposed PnP algorithm;
• 📙 Ablation_studies.ipynb: notebook display results from ablation studies;

Getting started 👾 🤖

To ensure reproducibility of the results, we suggest using the environment.yml file when creating an environment.

Simply run:

./install_env.sh

It will take some time, if at the end you see the word Done! on your terminal you are ready to go.

Remember to always activate the environment by typing:

conda activate ssdeblend

Disclaimer: All experiments have been carried on a Intel(R) Xeon(R) CPU @ 2.10GHz equipped with a single NVIDIA GEForce RTX 3090 GPU. Different environment configurations may be required for different combinations of workstation and GPU.