Python Package

pyAVS is the Python package for loading and preprocessing the AVS dataset – MEG, eye tracking, and structural MRI, ported from the analysis pipelines behind the AVS dataset paper into a documented, reusable library.

Three Steps: Configure, Compose, Analyze

  1. Configure – point pyAVS at your local copy of the dataset once, with pyavs.set_data_path() or the pyavs configure CLI command (see Installation).

  2. Compose – use AVSComposer (see AVSComposer Guide) to load MEG data, run ICA/filtering, and align eye-tracking events into epoched MEG data with rich metadata.

  3. Analyze – source-reconstruct, extract ROI/population codes, or hand epochs off to your own analysis code. See Tutorials and Examples for worked examples across preprocessing, source reconstruction, object detection, and configuration/reproducibility.

AVSComposer Guide

The recommended entry point for MEG + eye-tracking data fusion, with a full worked example and advanced configuration options.

AVSComposer Guide
Tutorials

Narrative, end-to-end walkthroughs: MEG + eye tracking, and source reconstruction + population codes.

Tutorials
Examples

Focused, runnable examples for source reconstruction, cross-session filters, configuration, object detection, and reproducibility.

Examples
API Reference

Full API documentation, organized by submodule (preprocessing, source, scenes, dataloader, and more).

API Reference