Call: SIMSPINE Benchmark Extensions and Baselines
Abstract
This collaboration call focuses on extending SIMSPINE with stronger baselines, broader evaluation settings, and reproducibility-first benchmark protocols for spine-aware motion understanding.
Scope
This call is for collaborators interested in extending the SIMSPINE benchmark with stronger methods, cleaner protocols, and better reproducibility artifacts.
The goal is not only to improve headline numbers, but to make comparisons fair, transparent, and useful for future work in spine-aware pose estimation.
Open Collaboration Tasks
- Add stronger 2D and 3D reference baselines under a unified training and evaluation protocol.
- Expand stress tests for cross-subject generalization, motion complexity, and failure-case analysis.
- Improve evaluation tooling, reporting templates, and reproducibility checklists.
- Contribute ablation studies around simulation choices and biomechanical constraints.
What Is Already Available
- Public SIMSPINE dataset release.
- Initial baseline code and benchmark setup.
- Paper-level description of annotation pipeline and biomechanical checks.
Related Papers and Resources
- SIMSPINE (CVPR 2026)
- SpinePose (CVPRW 2025)
- SIMSPINE Dataset (Hugging Face)
- SpinePose Inference Library
Collaboration Format
- Joint protocol design and benchmark planning.
- Shared implementation checkpoints with reproducibility reviews.
- Co-authored technical reports or paper submissions when outcomes are publication-grade.
If this scope matches your interests, contact me at mukh07@dfki.de with a short note about your background and preferred collaboration style.
Maintained by saifkhichi96 on GitHub.
The website is distributed under different open-source licenses. For more details, see the notice at the bottom of the page.