v0.12.0

v0.12.0

Bug fix for MC-RISE when threading is utilized, plus exemplar notebook demonstrating its use.

Updates / New Features

Examples

• Added the MC-RISE example notebook that demonstrates the MC-RISE saliency
  algorithm on a sample image from the GTSRB dataset.

Fixes

v0.11.0

v0.11.0

Improved linting, formating, and dependency management.

Updates / New Features

CI/CD

• Updated and applied ruff configuration.

Fixes

v0.10.0

v0.10.0

Massive overhaul to improve typing and linting to support the CDAO JATIC program,
as well as adding a new saliency algorithm called MC-RISE to the repertoire.

Updates / New Features

Code

• Updated hundreds of typing and linting needs.

• Added MC-RISE classification saliency algorithm.

Documentation

• Removed a deprecated badge from the README.

• Updated linting for documents.

• Added Maintainer_Guide_GitLab_to_GitHub.md to provide detailed guidelines.
  for synchronizing GitHub and GitLab repositories.

Dependencies

• Updated the minimum supported scipy to address vulnerability.

CI/CD

• Added check for release notes on GitLab.

• Added numerous GitLab yml files required.

Fixes

v0.9.2

v0.9.2

This is a quick hotfix to remove a pinned version of tqdm.

Updates / New Features

Fixes

• Removed an accidental tqdm pin and put it as a lower-bound.

v0.9.1

v0.9.1

Fixed a bug where if no detections were found in an image, then the generator would fail.

Updates / New Features

Documentation

• Removed a deprecated badge from the README.

Implementations

• Added a check to exit early if no detections were found in PerturbationOcclusion

• Added a check to exit early if no saliency maps were generated in GenerateObjectDetectorBlackboxSaliency

Fixes

v0.9.0

v0.9.0

Dropped support for Python 3.8 and addressed dependency vulnerabilities.

Updates / New Features

Python

• Dropped support for Python 3.8.

CI/CD

• Dropped scanning for Python 3.8 and changed default to Python 3.9.

Documentation

• ReadTheDocs now uses Python 3.9.

• ReadTheDocs now uses poetry install.

Fixes

Notebooks

• Added pandas to examples/MNIST_scikit_saliency.ipynb.

v0.8.3

v0.8.3

A quick hotfix to address scikit-learn compatability issues.

Updates / New Features

Fixes

• Increased scikit-learn lower bound to address CVE-2024-5206

v0.8.2

v0.8.2

A quick hotfix to resolve Python 3.12 numpy dependencies.

Updates / New Features

Fixes

• Added numpy pin ^1.26 for Python 3.12.

v0.8.1

v0.8.1

Addressed a few dependency vulnerabilities and updated the CI/CD to current versions.

Updates / New Features

CI/CD

• Updated to use checkout@v4 and cache@v4 instead of *@v2.

Fixes

Dependencies

• Fixed numpy dependency versions for downstream resolution.

• Jupyter notebooks now installed with notebook instead of jupyter.

• Increased the lower bound of tqdm to 4.66.3 to address CVE-2024-34062.

• Ran poetry update to update poetry.lock for vulnerability scanning.

v0.8.0

v0.8.0

Updated dependencies to support Python 3.12 along with a few other smaller bug fixes.

Updates / New Features

CI/CD

• Changed CodeCov version from v3 to v4.

• Add python 3.12 to test matrix

Docker

• Update Dockerfile install of poetry, and make separate & specific directory
  copies.

Utils

• Updated logging format of occlusion masking benchmark utility.

Fixes

Tests

• Fix various floating point equality comparisons to not use exact match.

• Fix random number usage from numpy to use np.random.default_rng.

• Fix error being masked in test_sal_on_coco_dets