ONCERCO UAV reflectance tools

oncerco-uav is a public Python package for extracting and modeling multi-angular multispectral UAV reflectance. It provides polygon-aware pixel extraction, spectral filtering, and RPV modeling for MicaSense-compatible five-band stacks.

Public package layout

• src/oncerco_uav/core/: configuration, validation, preprocessing, and search helpers.
• src/oncerco_uav/extract/: camera, raster, geometry, and polygon extraction.
• src/oncerco_uav/filter/: parquet loading and spectral-index filtering.
• src/oncerco_uav/modelling/: RPV fitting and modeling helpers.
• src/oncerco_uav/pipelines/: extraction, filtering, and RPV command modules.
• scripts/preprocessing/: generic RedEdge-P and orthorectification tools.

The package does not include project data, private analysis, research results, presentations, or machine-local configurations.

Install

python -m pip install -r requirements.txt
python -m pip install -e .

Run the pipelines

Copy examples/config_file_example.yml, set its input/output paths, and run:

python -m oncerco_uav.pipelines.extract_data --config my_config.yml
python -m oncerco_uav.pipelines.filtering --config my_config.yml
python -m oncerco_uav.pipelines.modelling --config my_config.yml --band band1

The same commands are available as uav-extract, uav-filter, and uav-rpv.

For RedEdge-P stacks the output order is Blue, Green, Red, Red edge, NIR. Reflectance stacks use uint16 values with reflectance = pixel / 32767.

Development

make test
make test-e2e
make coverage
make build

CI enforces at least 95% line coverage for the public oncerco_uav package. The current local baseline is 95.06% (220 non-e2e tests passed; the six e2e tests are run separately).

Use synthetic or user-provided data paths in tests. Do not commit raw imagery, large derived products, parquets, TIFFs, or local mount paths.