v25.08.00

🚨 Breaking Changes

• Remove CUDA 11 from dependencies.yaml (#887) @KyleFromNVIDIA
• stop uploading packages to downloads.rapids.ai (#883) @jameslamb

🐛 Bug Fixes

• Drop unused variable CUDA_MAJOR_VERSION (#907) @jakirkham
• Fix test failures due to deprecated Pillow Image.fromarray 'mode' parameter (#901) @grlee77
• Adding GH_TOKEN pass-through to summarize job (#891) @msarahan

📖 Documentation

• add docs on CI workflow inputs (#897) @jameslamb

🛠️ Improvements

• Drop CUDA 11 references from docs (#914) @jakirkham
• fix(docker): use versioned -latest tag for all rapidsai images (#908) @gforsyth
• remove cuspatial references (#906) @jameslamb
• Drop CUDA 11 from CI Scripts (#903) @AyodeAwe
• run: fix compatibility with PEP-0632 for supporting Python >=3.12 (#902) @ElieDeBrauwer
• Enhance GPU memory leak test with unified memory check and increased thresholds (#898) @gigony
• Use CUDA 12.9 in Conda, Devcontainers, Spark, GHA, etc. (#894) @jakirkham
• Remove nvidia and dask channels (#893) @vyasr
• refactor(shellcheck): fix all shellcheck warnings/errors (#890) @gforsyth
• refactor(conda): remove cuda11 conditionals from conda recipes (#889) @gforsyth
• Remove CUDA 11 from dependencies.yaml (#887) @KyleFromNVIDIA
• stop uploading packages to downloads.rapids.ai (#883) @jameslamb
• Forward-merge branch-25.06 into branch-25.08 (#880) @gforsyth
• Forward-merge branch-25.06 into branch-25.08 (#876) @gforsyth

[NIGHTLY] v25.10.00

🔗 Links

• Development Branch
• Compare with main branch

🐛 Bug Fixes

• Update fmt library to version 11.2.0 in CMake configuration (#917) @gigony

🛠️ Improvements

• add a copy of rapids-configure-conda-channels to the repo (#923) @jameslamb
• Update rapids-build-backend to 0.4.1 (#922) @KyleFromNVIDIA

v25.06.00

🛠️ Improvements

• use 'rapids-init-pip' in wheel CI, other CI changes (#881) @jameslamb
• Finish CUDA 12.9 migration and use branch-25.06 workflows (#879) @bdice
• Build and test with CUDA 12.9.0 (#877) @bdice
• Refactor: Enhance GPU Memory Leak Test for read_region (#874) @gigony
• Add support for Python 3.13 (#873) @gforsyth
• Download build artifacts from GitHub for CI (#870) @VenkateshJaya
• Add ARM conda environments (#867) @bdice
• Moving wheel builds to specified location and uploading build artifacts to Github (#854) @VenkateshJaya
• Region Properties Performance Overhaul - Part 6: Public API (regionprops_table) (#848) @grlee77
• Region Properties Performance Overhaul - Part 5: Perimeter and Euler Characteristic (#847) @grlee77
• Region Properties Performance Overhaul - Part 4: Moment-Based Properties (#846) @grlee77
• Region Properties Performance Overhaul - Part 3: Convex Image Properties (#845) @grlee77
• Region Properties Performance Overhaul - Part 2: Intensity Image Properties (#844) @grlee77
• Region Properties Performance Overhaul - Part 1: Basic Properties (#843) @grlee77

v25.04.00

🐛 Bug Fixes

• CuPy 13.4.1 compatibility: Fix dtype handling in fused chan-vese kernels (#856) @grlee77
• Fix path in update-version (#852) @raydouglass
• avoid potential CUDA out of bounds memory access in test case (#851) @grlee77
• Consistently raise error on non-CuPy input to regionprops functions (#849) @grlee77
• update vendored binary_fill_holes (#842) @grlee77

📖 Documentation

• Add cuCIM pronunciation to README (#823) @jakirkham

🛠️ Improvements

• Update various library versions to meet the minimum required CMake version (>= 3.5) for compatibility with CMake 3.30.4. (#858) @gigony
• Use conda-build instead of conda-mambabuild (#841) @bdice
• require sphinx<8.2.0 (#840) @jameslamb
• Improve performance of label2rgb (#839) @grlee77
• add utility for memory efficient maximum pairwise distance computation with GPU support (#838) @grlee77
• vendor CUDA-accelerated find objects (#837) @grlee77
• update vendored binary morphology code (#836) @grlee77
• Consolidate more Conda solves in CI (#835) @KyleFromNVIDIA
• Require CMake 3.30.4 (#834) @robertmaynard
• Create Conda CI test env in one step (#833) @KyleFromNVIDIA
• Add build_type input to test.yaml (#831) @gforsyth
• Implement convex_hull_image and convex_hull_object (#828) @grlee77
• Use shared-workflows branch-25.04 (#826) @bdice
• add telemetry (#822) @msarahan
• raise setuptools floor to 61.0.0, sphinx floor to 8.0.0, other small dependencies cleanup (#820) @jameslamb
• Forward-merge branch-25.02 to branch-25.04 (#819) @bdice
• Migrate to NVKS for amd64 CI runners (#817) @bdice

v25.02.00

🐛 Bug Fixes

• Fix primitives benchmark code (#812) @gigony
• CuPy 14.0 compatibility (#808) @grlee77

🛠️ Improvements

• Use rapids-pip-retry in CI jobs that might need retries (#824) @gforsyth
• Revert CUDA 12.8 shared workflow branch changes (#818) @vyasr
• Build and test with CUDA 12.8.0 (#815) @bdice
• Add shellcheck to pre-commit and fix warnings (#814) @gforsyth
• Update vendored ndimage code with axes support (#813) @grlee77
• Use GCC 13 in CUDA 12 conda builds. (#811) @bdice
• Improve performance of color distance calculations by kernel fusion (#809) @grlee77
• Incorporate upstream changes from scikit-image 0.25 (#806) @grlee77
• Update version references in workflow (#803) @AyodeAwe
• Require approval to run CI on draft PRs (#798) @bdice
• Add breaking change workflow trigger (#795) @AyodeAwe

v24.12.00

🚨 Breaking Changes

• Deprecations: carry out removals scheduled for release 24.12 (#786) @grlee77

🛠️ Improvements

• allow path conflicts in conda builds (#801) @jameslamb
• enforce wheel size limits, README formatting in CI (#797) @jameslamb
• build wheels without build isolation, print sccache stats in builds (#792) @jameslamb
• make conda installs in CI stricter (#791) @jameslamb
• Deprecations: carry out removals scheduled for release 24.12 (#786) @grlee77

v24.10.00

🐛 Bug Fixes

• Use cupy to measure memory leak (#777) @bdice
• Fix wheel tests for Rocky Linux 8. (#774) @bdice
• Disable custom 2D separable filtering kernels on windows (#770) @grlee77
• chan_vese: pass all constants to _fused_variance_kernel2 as device scalars (#764) @grlee77
• Fix "compatibility" spelling in CHANGELOG (#759) @jakirkham
• Fix error in dependencies.yaml causing incomplete pyproject.toml generation (#757) @grlee77

🛠️ Improvements

• Do not depends on unused libraries for libtiff (#785) @gigony
• Fix a couple of performance issues in peak_local_max (improve performance of blob detectors and corner_peaks) (#782) @grlee77
  • blob_log, blob_dog and blob_doh should see performance improvement from this
  • peak_local_max and corner_peaks will see substantial performance improvement when min_distance=1. A smaller improvement should be seen for larger min_distance
• update vendored CUDA includes to match CuPy >= 13.3 (#781) @grlee77
• Use CI workflow branch 'branch-24.10' again (#780) @jameslamb
• Add support for Python 3.12 (#773) @jameslamb
• Update rapidsai/pre-commit-hooks (#772) @KyleFromNVIDIA
• bump scikit-image upper bound (allow 0.24.x) (#769) @grlee77
• Drop Python 3.9 support (#766) @jameslamb
• Remove NumPy <2 pin (#762) @seberg
• Update pre-commit hooks (#760) @KyleFromNVIDIA
• Switch to pytest-lazy-fixtures (#756) @gigony

v24.08.00

🛠️ Improvements

• Drop NumPy build dependency (#751) @jakirkham
• Use workflow branch 24.08 again (#749) @KyleFromNVIDIA
• Build and test with CUDA 12.5.1 (#747) @KyleFromNVIDIA
• Minor fixes for NumPy 2.0 compatiblity (#746) @grlee77
• skip CMake 3.30.0, require CMake >=3.26.4 (#745) @jameslamb
• Use verify-alpha-spec hook (#744) @KyleFromNVIDIA
• remove .gitattributes (#740) @jameslamb
• Adopt CI/packaging codeowners (#739) @bdice
• Remove text builds of documentation (#738) @vyasr
• use rapids-build-backend (#736) @jameslamb

v24.06.00

🚨 Breaking Changes

• The output argument of cucim.skimage.filters.gaussian has been renamed to out. The old name is deprecated and will be removed in release 25.02 (#727)
• Renamed get_xyz_coords function is now removed (use skimage.color.xyz_tristimulus_values instead) (#724)
• Removed deprecated return_error kwarg from phase_cross_correlation (the error is now always returned) (#724)
• Removed deprecated random_state kwarg from medial_axis (it was renamed to rng previously) (#724)

🐛 Bug Fixes

• Use SciPy's KDTree instead of deprecated cKDTree (#733) @grlee77
• Binary and grayscale morphology functions have bug fixes in the case of even-sized/non-symmetric footprints (for details see upstream MR: scikit-image/scikit-image#6695) (#728)

🚀 New Features

• cucim.skimage.measure.regionprops (and regionprops_table) support one new region property: intensity_std (#727)
• cucim.skimage.segmentation.expand_labels now supports a spacing keyword argument to take a pixel's physical dimensions into account (#727)
• binary morphology functions have a new mode argument that controls how values outside the image boundaries are interpreted (#728)
• grayscale morphology functions have new mode and cval arguments that control how boundaries are extended (#728)

🛠️ Improvements

• Enable FutureWarnings/DeprecationWarnings as errors (#734) @mroeschke
• Migrate to {{ stdlib("c") }} (#731) @hcho3
• Implement upstream changes from scikit-image 0.23 (part 2 of 2: morphology) (#728) @grlee77
• Implement upstream changes from scikit-image 0.23 (part 1 of 2) (#727) @grlee77
• Update the test criteria for test_read_random_region_cpu_memleak (#726) @gigony
• Remove code needed to support Python < 3.9 and apply ruff's pyupgrade rules (#725) @grlee77
• removal of deprecated functions/kwargs scheduled for release 24.06 (#724) @grlee77
• Enable all tests for arm jobs (#717) @galipremsagar
• prevent path conflict (#713) @AyodeAwe
• Updated cuCIM APIs for consistency with scikit-image 0.23.2 (#727 and #728)
• Additional modules use __init__.pyi instead of just __init__.py (#727)
• Some grayscale tests now compare directly to skimage CPU outputs instead fetching previously saved values (#728)
• Refactored some test cases to better use pytest.mark.parametrize (#728)
• Bumped version pinning for scikit-image to allow 0.23.x to be installed (#728)

📖 Documentation

• Various fixes to documentation strings (consistent shape notation, etc.) (#727)

cuCIM 24.04.00 (10 Apr 2024)

🐛 Bug Fixes

• Require click as a wheel dependency (#719) @jakirkham
• Fix docs upload directory (#714) @raydouglass
• Fix popd indent in run (#693) @jakirkham
• Re-run ci/release/update-version.sh 24.04.00 (#690) @jakirkham

🚀 New Features

• Support CUDA 12.2 (#672) @jameslamb

v24.04.00

🐛 Bug Fixes

• Require click as a wheel dependency (#719) @jakirkham
• Fix docs upload directory (#714) @raydouglass
• Fix popd indent in run (#693) @jakirkham
• Re-run ci/release/update-version.sh 24.04.00 (#690) @jakirkham

🚀 New Features

• Support CUDA 12.2 (#672) @jameslamb

🛠️ Improvements

• Use conda env create --yes instead of --force (#716) @bdice
• Add upper bound to prevent usage of NumPy 2 (#712) @bdice
• Remove hard-coding of RAPIDS version (#711) @KyleFromNVIDIA
• Switch pytest-xdist algorithm to worksteal (#708) @bdice
• Simplify version update script (#705) @jakirkham
• Add support for Python 3.11, require NumPy 1.23+ (#704) @jameslamb
• target branch-24.04 for GitHub Actions workflows (#702) @jameslamb
• Refactor CUDA libraries in dependencies.yaml (#699) @bdice
• Update ops-bot.yaml (#694) @AyodeAwe
• add rapids-dependency-file-generator pre-commmit hook (#682) @jameslamb