CLI

CamCalib installs a single entry point, camcalib, with seven subcommands. Everything is config-driven — flags select what to run, the YAML says how.

camcalib --help

Subcommand	Purpose
calibrate	Run calibration
cam	Camera-only calibration from a CamRunConfig YAML
cam-imu	Joint camera–IMU calibration
apply-calib	Apply a calibration to a raw bag → corrected bag
roundtrip	calibrate → apply-calib → recalibrate → PASS/FAIL report
run	Run calibration, dispatching on the config type
analyze	Analyse calibration results

Hand-written, and therefore able to drift

Generating this page from the parsers is on the roadmap. Until then camcalib <subcommand> --help is authoritative.

calibrate

Camera calibration from a config file.

camcalib calibrate --config config.yaml

Flag	Description
--config, -c	Path to the calibration config YAML

Config schema: Configuration.

cam

Camera-only calibration driven by a CamRunConfig YAML — the run-oriented variant, used by the batch and regression harnesses.

camcalib cam --config example_cam_run.yaml

Flag	Description
--config, -c	Path to the CamRunConfig YAML

cam-imu

Joint camera–IMU calibration: extrinsics T_CtoI, time offset td, rolling-shutter readout, IMU intrinsics, and biases.

camcalib cam-imu --config cam_imu.yaml

Flag	Description
--config, -c	Path to the cam-IMU config YAML
--bag	Override the input bag from the config

List several sequences under input.bags and each is calibrated into its own discoverable bag<NN>/cam_imu.json — a whole dataset family from one config. See Camera–IMU calibration.

apply-calib

Apply a saved cam-IMU calibration back onto a raw ROS1 bag and re-record it calibration-corrected.

camcalib apply-calib -c recalib.yaml
camcalib apply-calib --calib cam_imu.json --in raw.bag --out corrected.bag

Flag	Description
-c, --config	Config bundling calib / input / output
--calib	Calibration result to apply
--in	Raw input bag
--out	Output bag; defaults to <bag>.corrected.bag
--map	Explicit topic→sensor map. Rarely needed — the calib records each sensor’s source topic, so the mapping is auto-derived
--keep-bias	Do not remove the static bias
--undistort	Also undistort images

What it corrects and why: Corrected-dataset export.

roundtrip

End-to-end validation: calibrate, apply the calibration to its own bag, recalibrate the corrected bag, and report PASS/FAIL. Exits non-zero on FAIL, so it works in CI.

camcalib roundtrip -c cam_imu.yaml

Flag	Description
-c, --config	Cam-IMU config to validate
--out	Report output path
--bag	Override the input bag
--keep-corrected	Keep the corrected bag (auto-deleted otherwise)
--keep-bias	Pass through to apply-calib

What PASS means: Round-trip validation.

run

Runs a calibration, dispatching on the config type so you do not have to pick the subcommand yourself.

camcalib run --config any_config.yaml

Flag	Description
--config, -c	Any CamCalib config

analyze

Analyse existing calibration results.

camcalib analyze --results results/regression/dataset/run/

Flag	Description
--results, -r	Results directory to analyse

Scripts

Not subcommands, but part of the everyday surface:

Script	Purpose
scripts/run_regression_tests.py	Manifest-driven batch runner. --level benchmark --parallel 8, --only <dataset>
scripts/run_quick_regression.py	Six-run subset, roughly ten minutes
scripts/compare_baseline.py	Compare current results against a saved baseline
scripts/save_baseline.py	Save a new baseline, with a comparison report
scripts/visualize_result.py	Post-hoc Rerun 3D viewer for a result directory
python -m camcalib.analysis.regenerate_html	Rebuild HTML reports from saved YAML + detection cache
studies/calib_video_real.py	Render a calibration MP4 (--sim, --recover, --rig-view)

More detail in Evaluation & metrics.