Tags: talmolab/sleap-io
Tags
chore(release): Bump version to 0.9.2 (#553) * chore(release): Bump version to 0.9.2 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * style: fix ruff format trailing-comma violations in codecs/README.md Unrelated pre-existing formatting drift surfaced by CI's Lint job on this PR (first PR since #549 to touch code paths and trigger ruff format --check). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
chore(release): Bump version to 0.9.1 (#544) Claude-Session: https://claude.ai/code/session_011DiaqkmRcp1L9yD8dx4jLj Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
chore(release): Bump version to 0.9.0 (#541) Claude-Session: https://claude.ai/code/session_011DiaqkmRcp1L9yD8dx4jLj Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
fix(io): preserve isolated-node order in standalone skeleton JSON (L3) ( #512) The jsonpickle skeleton decoder built the final node list only from py/id back-references in the nodes section, so isolated (edge-less) nodes — emitted inline as full py/object entries — were missed, tripping the "natural order" fallback that appends them at the tail (e.g. [iso, a, b] decoded to [a, b, iso]). Resolve both py/id refs and inline py/object nodes (by name, to the canonical decoded object) so the declared nodes-section order is preserved. The .slp path was already unaffected; #438 had already stopped the prior node-dropping. Claude-Session: https://claude.ai/code/session_01A6ZBs7Yww49sCmFF6XSSHs Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
chore(release): Bump version to 0.7.1 (#437) Includes small follow-ups surfaced by the v0.7.1 release review: - docs(cli): Document --trails / --trail-* flags under `sio render` (Quick Reference + Motion Trail Options subtable). Closes the only CLI doc gap from #434. - docs(model/labels): Cross-reference Labels.match_video and the widened lookup signatures introduced in #436 from the Querying section. - docs(matching): Refresh ConflictResolution.conflict_type docstring to cite the actually-emitted values ("instance_conflict", "negative_flag_conflict" from #432) instead of the stale "duplicate_instance" / "skeleton_mismatch" examples. - docs(codecs): Note that to_analysis_arrays is an internal building block behind save_analysis_h5 (it is intentionally omitted from sleap_io.codecs.__all__). Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
fix(io/tiff): suppress ambiguous-pages warning when content rules out… … class-stack (#422) Multi-page TIFFs with no axis metadata previously always warned "Assuming pages are time... pass pages_as='classes'", even when pages had multiple distinct positive label IDs per page. That layout cannot be a stack of binary class masks, so the suggested alternative would destructively collapse instance IDs via from_binary_masks. Gate the warning on _pages_could_be_class_stack(): inspect each loaded page and suppress the warning if any page has >1 unique positive value. Purely binary multi-page TIFFs remain genuinely ambiguous and still warn. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
fix: Preserve ROIs/masks in lazy copy and fix COCO empty segmentation… … fallback (#361) - Fix lazy Labels.copy() to include rois and masks (previously omitted, causing data loss when copying lazy-loaded labels with ROI/mask annotations) - Fix COCO reader to fall back to bbox ROI when segmentation is an empty list (previously only checked for None, missing the [] case) - Add tests for both edge cases Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
PreviousNext