FiLark Documentation

Distributed acoustic sensing
Streaming-first DAS visualization, annotation, and analysis integration.
FiLark keeps large DAS recordings navigable without forcing every workflow through short pre-extracted snippets. The GUI is the interactive front end; the I/O, DSP, pipeline, and visualization layers remain usable from scripts.
Start with setup Read architecture

Public Release Note

This public package is a staged preview. Core I/O, visualization, ROI analysis, signal processing, and comparison workflows are available now. Annotation and monitoring GUI workflows are visible for preview and will be released with the complete 0.0.4 version after the FiLark paper is published.

Demo Videos

The following videos show FiLark workflows from interactive browsing to annotation, monitoring, and review:

Exploration and Processing Bilibili backup
Large Data and Annotation Bilibili backup
Monitoring and Recheck Bilibili backup

Choose A Path

Getting Started

Install FiLark, launch the GUI, and open a first data source.

Getting Started

Basic Usage

Walk through loading, navigation, ROI analysis, annotation, monitoring, and recheck.

Basic Usage

Concepts

Understand the streaming-first design, package boundaries, and coordinate model.

Concepts

Guides

Task-oriented pages for loading, visualization, annotation, monitoring, DSP, and plugins.

Guides

Showcase Workflows

Read how the pieces fit together in common DAS analysis sessions.

Showcase Workflows

Core Capabilities

Streaming I/O

H5, NPY, binary, array, and folder-backed sources are exposed through one metadata-aware Tape interface.

Bounded-Memory Viewer

The main canvas uses a fixed-size GPU texture and updates only newly visible regions while navigating.

Display Processing

Attach lightweight stream preprocessing and display filters for exploratory browsing.

Coordinate-Stable Labels

Annotations and monitor events are stored in time-channel coordinates, not display pixels.

Algorithm-In-The-Loop

Monitor plugins connect external algorithms to the viewer while the interface remains responsive.

ROI Analysis

Open PSD, STFT, F-K, tau-p, RMS, SVD, xcorr, and denoising views from selected regions.

Headless Processing

Use filark.io, filark.dsp, and filark.pipeline without starting Qt.

Project Notes

This site is a practical manual for the current repository. It borrows the project motivation from the FiLark paper, but it is written for people who need to run, configure, extend, or maintain FiLark.