Started as class project for CS230 at Stanford Univeristy, Spring 2023.

This is a fork of the original.

Convolutional Autoencoder for Compressing Distributed Acoustic Sensing Data from Urban Environments

The Autoencoder model is based on Lossy Data Compression from Tensorflow's official tutorial: https://www.tensorflow.org/tutorials/generative/data_compression

Project Description

Training an autoencoder to compress vast Distributed Acoustic Sensing (DAS) data. The goal is to achieve data compression with minimal loss.

Setup and Dependencies

Code has been tested in Google Colab on T4, L4, V100, and A100 instances.

Required dependencies:

1. Python (version 3.10.11)
2. TensorFlow (version 2.12.0)
3. NumPy (version 1.22.4)
4. Matplotlib (version 3.7.1)

Dataset

The data used in this project can potentially be obtained by contacting tculliso-[AT]-stanford-dot-edu

Authors

Haipeng Li

Thomas Cullison

Hassan Almomin

Department of Geophysics, Stanford University, 2024