v0.1.0
This is the initial release of this repository which hosts integration documentation, examples, and code related to integrating nrtk components with the JATIC needs and use-cases.
Updates / New Features
Documentation
Added reference to original NRTK documentation.
Added ability to render documentation on Gitlab Pages.
Update README.md
Added Sphinx document rendering for MRs. The docs pages can be accessed by clicking the “View App” button located in the merge request page under the test pipeline section.
Added sphinx auto-documentation on JATIC interoperability for object detection.
Added sphinx auto-documentation on JATIC interoperability for image classification.
Added email to pyproject.toml
Examples
Added an example notebook exploring the current state of several augmentation tools as well as the usability of the JATIC augmentation protocol.
Added an example notebook exploring the DPDivergence metric using the DAML tools developed by the ARIA team.
Added an example notebook exploring using Gradio as an interface for applying pyBSM perturbations.
API
Added in REST API server which accepts a POST request with a JSON data payload and returns a JSON payload
Added in unit-test for POST request with REST API server
Added JSON schema pydantic file and corresponding documentation for how it should be used
Added additional POST request specifically to accept the AUKUS json schema and call our main API with that data
Added
Dockerfileandcompose.yamlto containerize thenrtk-cdaopackage and runnrtk-perturber-cli.Added config file support for loading pyBSM factories for NRTKaaS.
License
Add Apache 2.0 license
Interoperability
Added
JATICDetectionAugmentationwhich performs perturbations on MAITE-protocol based object detection datasets.Added
COCOJATICObjectDetectionDataset`which loads a COCO detection dataset from disk and converts it to a MAITE-protocol compliant detection dataset.Added
JATICObjectDetectionDatasetwhich is a custom I/O bridging dataset wrapper for MAITE compliant detection datasets.Added a utility function to save a MAITE object detection dataset to file as a COCO dataset.
Added
JATICClassificationAugmentationwhich performs perturbations on MAITE-protocol based image classification datasets.
Utils
Moved kwcoco to poetry tools dependency
Added a CLI script to perform perturbations on a sample image set.
Added
maite-cpuextras option for CPU version of MAITE/torch
Python Version Support
Added support for
py3.12
CI/CD
Updated CI test matrix to support
py3.12Added additional entrypoint testing.
Major overhaul of pipeline to improve efficiency and yml readability.
Added
publishon tag toPyPi.Updated the
docsbuilding script to run properly onmain.
Other
Refactored package into ‘src/nrtk_jatic instead of ‘nrtk_jatic’
Saved a repeated return value in tests/api/test_app.py to a variable to save space
Fixes
Updated to use
nrtk>=0.5.3which patched an issue withnumpydependency resolution.Numpy is used by the package but was never added to the list of dependencies and was instead installed indirectly from
nrtk. Addednumpyas a required dependency, which also has the side effect of solving resolution issues.Updated
numpyhinge forPython 3.12Updated git lfs to properly track large files in any directory.