Open Source

Open Source

Independent software work focused on inspectable research workflows and practical computer-vision tools.

Alpha · Actively Developed

medlint

medlint is an alpha-stage, evidence-first Python library for auditing declared medical-imaging dataset splits before model training. It examines exact file relationships and selected DICOM identifiers across splits, then produces a deterministic report that separates observed evidence from audit coverage. The workflow is local, read-only, and designed to make potential contamination risks reviewable without changing source data. Its public API currently serves researchers and engineering teams that need a compact, inspectable pre-training checkpoint.

  • PyPI v0.1.1
  • Python 3.10–3.13
  • CI with strict typing and tests
  • Deterministic JSON report and published schema

Scope Current checks focus on byte-identical files and selected DICOM identifiers across declared dataset splits. Review the full scope on GitHub.

BiyoVes

BiyoVes is an open-source Python library and browser-based Web Studio for standards-based biometric, passport, and visa photo preparation. It combines SCRFD-based face detection and alignment with portrait matting, format-specific crop geometry, and automated quality preflight. The workflow supports standard photo formats and print templates, including 300-DPI PDF output, while preflight warnings remain advisory rather than a guarantee of issuing-authority acceptance. The library and Web Studio provide two access paths to the same preparation workflow.

  • PyPI v1.4.1 · Python 3.8–3.13
  • Browser-based Web Studio
  • CI, Pages, and release workflows
  • 300-DPI PDF output

View all repositories on GitHub