-
Create the torchvideo environment:
$ conda env create -n torchvideo -f environment.yml $ conda activate torchvideo
-
Check everything is installed properly by running the tests:
$ make test
-
Check you can build the documentation
$ make docs # For linux: $ xdg-open docs/build/html/index.html # For macOS: $ open docs/build/html/index.html
-
Set up pre-commit hooks:
$ pip install pre-commit $ pre-commit install
These will run every time you commit and ensure the code type checks, doesn't have trailing whitespace, runs the black formatter etc.
- Ensure docstrings are Google style
- Ensure changes have automated tests
- Ensure there are documentation updates if necessary
- Add changes to
CHANGELOG.md
- Implement your dataset with tests in
src/torchvideo/datasets
. - Add a new entry
docs/source/datasets.rst
of the form:MyDataset ~~~~~~~~~ .. autoclass:: MyDataset :special-members: __getitem__,__len__
- Add an example of usage to
examples/datasets.ipynb
- Implement your transformation with tests in
src/torchvideo/transforms
, ideally splitting it into a pure functional core, and a class that calls this. - Add a new entry
docs/source/transforms.rst
of the form:MyTransform ~~~~~~~~~~~ .. autoclass:: MyTransform :special-members: __call__
- Add an example of usage to
examples/transforms.ipynb
- Implement your sampler with tests in
src/torchvideo/samplers
. - Add a new entry
docs/source/samplers.rst
of the form:MySampler ~~~~~~~~~ .. autoclass:: MySampler
- Add an example of usage to
examples/samplers.ipynb