Skip to content

v2.0.1

Choose a tag to compare

View all tags
@github-actions github-actions released this 27 Apr 22:09
· 8 commits to main since this release
v2.0.1
34d44ac

CoastSeg v2.0.1

Release Date: 4/27/26

Minor release to fix attributions on Zenodo.

Highlights

  • Updated download tide model notebook instructions

    • The notebook instructions now correctly explain the clipping process. #321

  • Overhauled zoo workflow

    • CoastSeg 2.0 redesigns the zoo workflow by removing the optional TensorFlow, Transformers dependencies from the main CoastSeg environment.
    • Code to run the Zoo segmentation models now lives in a dedicated segmentation_workflow subfolder, where users can install the smaller dependency set with pixi install or a separate Conda environment.
    • The workflow still works the same from the user’s perspective, but segmentation now runs in a separate subprocess for a cleaner, more maintainable architecture.
    • The segmentation process has also been streamlined for a faster experience.
    • Removed the TESTTIMEAUG and otsu controls from the interface to simplify the workflow. #294
    • Zoo models are now downloaded from Hugging Face instead of Zenodo and saved to CoastSeg/models. #272

  • ONNX support for model inference

    • Added onnxruntime, allowing ONNX models to run directly in the CoastSeg environment.
    • The good/bad filtering models now use ONNX format, removing their TensorFlow dependency and making them easier to use and maintain.

New in the zoo workflow

  • Two options for extracting shorelines from segmentations
    • The zoo workflow now supports extracting shorelines either from a folder of existing segmentations or by running the model and extracting shorelines as part of the same workflow.
    • This gives users more flexibility in how they use the segmentation (zoo) workflow, whether they prefer working in a Pixi environment or a Conda environment.
zoo_two_workflow_options_extract_shorelines
  • Preview uploaded shorelines and transects before extraction
    • Users can now preview the shorelines and transects they upload before extracting shorelines.
    • This makes it easier to confirm that the correct inputs are loaded before running the extraction step.
zoo_preview_uploaded_features

Other Updates

  • Removed dependency on doodleverse_utils

    • CoastSeg no longer depends on doodleverse_utils, helping simplify the environment and reduce overhead.

  • Broader Python compatibility

    • CoastSeg is now compatible with more Python versions, making installation and setup easier across different systems.

Resolved Issues

Contributors

  • @2320sharon for the refactor, documentation improvements, download/CI fixes, and UI updates.
  • @Cjanda12 for the notebook UI suggestion that inspired the new bordered controls.
    -@mlundine for help streamlining the zoo workflow
    -@ddbuscombe for creating and providing the zoo segmentation models

Full Changelog

Full Changelog: v1.7.2...v2.0.1

Contributors

ddbuscombe, 2320sharon, and 2 other contributors
Assets 2
Loading