classtree

Classtree is a hierarchical classifier for images or text.

https://github.com/aiml-au/classtree

Science Score: 44.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • CITATION.cff file
    Found CITATION.cff file
  • codemeta.json file
    Found codemeta.json file
  • .zenodo.json file
    Found .zenodo.json file
  • DOI references
  • Academic publication links
  • Academic email domains
  • Institutional organization owner
  • JOSS paper metadata
  • Scientific vocabulary similarity
    Low similarity (9.8%) to scientific vocabulary

Keywords

classification hierarchical-classification machine-learning
Last synced: 4 months ago · JSON representation ·

Repository

Classtree is a hierarchical classifier for images or text.

Basic Info
Statistics
  • Stars: 2
  • Watchers: 2
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Topics
classification hierarchical-classification machine-learning
Created about 2 years ago · Last pushed over 1 year ago
Metadata Files
Readme License Citation

README.md

Classtree

Classtree is a hierarchical classifier for images or text.

shell pip install classtree

The fastest way to use Classtree is to call the CLI on a folder of images or text files.

shell train_data/ |- animals/ |- mammals/ |- marsupials/ |- koala/ |- image001.jpg |- image002.jpg |- ... |- ... |- reptiles/ |- ... |- ... |- ...

shell classtree train images --model animals --dir train_data/animals

or

shell classtree train text --model animals --dir train_data/animals

And then use your model with the predict command.

```shell classtree predict --model animals new_data/image304.jpg

birds/raptors/eagle ```

Pre-trained Models

You can download a pre-trained model using the download command.

shell classtree download --model dbpedia

Or download a pre-prepared dataset.

shell classtree download --text dbpedia

If you want to fine-tune an existing model, you can use the --from flag during training with any downloaded model.

shell classtree train text --model animals --from dbpedia --dir train_data/animals

Available Models

| Task | Name | Size | Dataset | Notes | |----------------------|--------------------|------|------------------------|----------------------------------------------| | Text Classification | dbpedia | M | dbpedia | |

Available Datasets

| Type | Name | Dataset | Notes | |-------|--------------------|------------------------------------------------------------------------------------|----------------------------------------------| | Image | inaturalist21-mini | iNaturalist 2021 (Mini) | Non-commercial research/educational use only | | Text | dbpedia | DBPedia | CC0: Public Domain |

Evaluation

You can test your model on a hold-out dataset using the test command.

shell classtree test --model animals --dir=test_data/animals

Licensing

Classtree is available for non-commercial internal research use by academic institutions or not-for-profit organisations only, free of charge. Please, see the license for further details. To the extent permitted by applicable law, your use is at your own risk and our liability is limited. Interested in a commercial license? For commercial queries, please email aimlshop@adelaide.edu.au with subject line “Classtree Commercial License”.

This is an AIML Shop project.

Owner

  • Name: Australian Institute for Machine Learning
  • Login: aiml-au
  • Kind: organization
  • Location: Australia

Citation (CITATION.cff)

cff-version: "1.2.0"
title: "Classtree"
message: "If you use this software, please cite both the article from preferred-citation and the software itself."
authors:
    - family-names: Valmadre
      given-names: Jack
    - name: AIML
preferred-citation:
  authors:
    - family-names: Valmadre
      given-names: Jack
  title: "Hierarchical classification at multiple operating points"
  type: inproceedings
  year: 2022
  booktitle: "Advances in Neural Information Processing Systems"
  volume: 35
  start: 18034
  end: 18045
  publisher: "Curran Associates, Inc."
  url: "https://proceedings.neurips.cc/paper_files/paper/2022/file/727855c31df8821fd18d41c23daebf10-Paper-Conference.pdf"
  editors:
    - name: "S. Koyejo and S. Mohamed and A. Agarwal and D. Belgrave and K. Cho and A. Oh"

GitHub Events

Total
  • Watch event: 2
Last Year
  • Watch event: 2

Dependencies

.github/workflows/publish.yml actions
  • actions/checkout v2 composite
  • actions/download-artifact v2 composite
  • actions/setup-python v2 composite
  • actions/upload-artifact v2 composite
pyproject.toml pypi
  • Pillow *
  • fsspec *
  • gcsfs *
  • matplotlib *
  • numpy *
  • torch *
  • torchtext *
  • torchvision *
  • tqdm *
requirements.txt pypi
  • Flake8-pyproject *
  • bandit *
  • black *
  • flake8 *
  • mypy *
  • pre-commit *