Skip to content

Releases: AndreaF-UniBO/PyWALL

Release list

PyWALL v13 (0.13.0)

PyWALL v13 (0.13.0) Pre-release
Pre-release

Choose a tag to compare

View all tags
@AndreaF-UniBO AndreaF-UniBO released this 03 Aug 15:54
v0.13.0
4ac0cec
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

PyWALL v13 is the first public release under the PyWALL name. It remains alpha-quality research software: segmentation outputs require expert archaeological review.

Highlights

  • K-Means masonry-image segmentation remains available as the self-contained baseline.
  • SAM 2 now uses only Meta's official facebookresearch/sam2 backend.
  • The historical Segmentazione DL control and TopoMortar/Ultralytics runtime have been removed.
  • PNG mask and scaled DXF contour exports are retained.
  • Reproducible Windows setup, checkpoint download, launch and test scripts are included.

Installation assets

Model weights, virtual environments and test photographs are not included. Run scripts/setup_windows.ps1 and scripts/download_sam2_checkpoint.ps1 from a source checkout; the checkpoint download is verified by SHA-256.

Verification

  • GitHub Actions passed on Python 3.11 / Windows.
  • Local core, GUI, K-Means, PNG and DXF tests passed.
  • A reduced real inference test passed with the official Meta SAM 2 backend on an NVIDIA RTX 4060 Laptop GPU.
  • The user confirmed the local interactive application functions correctly.

See the repository README and docs/local-testing.md for limitations and exact verification scope.

Loading

WALLpy 0.1.0

WALLpy 0.1.0 Latest
Latest

Choose a tag to compare

View all tags
@AndreaF-UniBO AndreaF-UniBO released this 03 Aug 09:23
v0.1.0
f041a56

WALLpy 0.1.0

This is the first public release of WALLpy, a preliminary desktop research application for masonry-image segmentation and archaeological drawing support.

Included

  • K-Means segmentation using colour, edge, and local-texture features.
  • Optional SAM 2 integration through separately installed backends and weights.
  • Morphological and colour-based refinement, manual erasing, and contour review.
  • Binary PNG-mask and scaled DXF export.
  • Installable Python metadata and the wallpy console command.
  • Apache-2.0 license, citation metadata, third-party notices, tests, and CI.

Validation

  • Tested on Windows with Python 3.11.
  • Clean-environment installation and wheel installation verified.
  • Seven deterministic unit and smoke tests passed.
  • The desktop interface was launched successfully without loading archaeological data.

Important limitations

  • Results require expert archaeological review and are not autonomous interpretations.
  • The legacy supervised model, datasets, SAM checkpoints, and other model weights are not distributed.
  • The optional Ultralytics backend is separately licensed under AGPL-3.0.
  • Operating systems other than Windows and Python versions other than 3.11 have not yet been verified.

See the repository README for installation, usage, and troubleshooting instructions.

Assets 3
Loading