| # Dependencies for the OpenH-RF eSAF sub-dataset package. |
| # Install with: python -m pip install -r requirements.txt |
| # |
| # reconstruct.py follows the OpenH-RF example templates and runs the processing |
| # chain defined in pipeline.yaml through the zea.Pipeline — zea (>= 0.1.0, the |
| # version the files were written with via zea.File.create) and one Keras backend |
| # (jax below; tensorflow/torch also work, set KERAS_BACKEND accordingly) are |
| # required. Run with: KERAS_BACKEND=jax python reconstruct.py |
| numpy>=1.24 |
| scipy>=1.10 |
| matplotlib>=3.6 |
| h5py>=3.8 |
| pyyaml>=6.0 |
| zea>=0.1.0 |
| jax |
|
|
| # Notes: |
| # - verify_submission.py's structural checks (deliverables, HDF5 schema, manifest, |
| # pipeline.yaml) need only numpy/h5py/pyyaml; its reconstruction smoke test runs |
| # reconstruct.py and therefore needs zea + a backend like everything else. |
| # - extras/esaf_demo.py (optional eSAF demonstration) needs only numpy/scipy/ |
| # matplotlib/h5py. |
|
|
