Movement Anywhere: An Open-Source Distributed 2D Video-Based Movement Analysis Platform Empowered by Active Learning.
Where this comes from
- Record sourced from PubMed, PMID 42090537.
- Also identified by DOI 10.1109/JBHI.2026.3690720.
- No licence information is recorded for this record.
- Because redistribution is not established, this page shows the abstract only. Follow the links below for the full text.
Abstract
Movement analysis plays a pivotal role in diagnosing and monitoring neurodegenerative and musculoskeletal diseases. Traditional tools, such as 3D motion capture systems and electronic walkways, though effective, are costly and spatially demanding, limiting their accessibility. To address these challenges, we introduce Movement Anywhere, an open-source, distributed plat form for 2D video-based movement analysis. This plat form features advanced tracking algorithms that robustly handle scenarios where patients require assistance, and it rigorously establishes the conditions necessary for precise depth information extraction, crucial for accurate motion parameter estimation. Additionally, Movement Anywhere is adaptable to various 2D cameras and in corporates an active learning framework to streamline algorithm updates. Evaluated with datasets from multiple medical centers, our approach demonstrates substantial improvements over previous methods. Movement Anywhere provides a cost-effective, scalable, extensible, and user-friendly solution for effective disease monitoring and progression tracking. Movement Anywhere is accessible at https://movement-anywhere.cmdm.tw/, with source codes provided at https://github.com/ Kaminyou/Movement-Anywhere.