Installation

CamCalib is pure Python and installs with pip on Windows, macOS, and Linux. There is no ROS dependency and nothing to compile.

One-step setup (recommended)

bash scripts/setup_env.sh
conda activate cam_calib

Creates the cam_calib conda environment, installs every dependency, and runs the test suite so you know the install works before you point it at data.

Manual install

pip install -e .

Detector backends are optional extras — install the ones your boards need:

pip install -e ".[apriltag]"    # AprilGrid (aprilgrid package)
pip install -e ".[aruco]"       # ArUco (opencv-contrib)
pip install -e ".[dev]"         # pytest

Reading bags

ROS bags are read through rosbags — pure Python, no ROS installation:

pip install rosbags

This is what lets a ROS1 or ROS2 bag be calibrated on a machine that has never had ROS on it.

GTSAM needs NumPy 1.x

GTSAM 4.2 from pip is built against the NumPy 1.x ABI. With NumPy 2 installed you will get a segfault when constructing gtsam.Pose3(Rot3, np.array) — not an exception, a crash. Pin it:

pip install "numpy<2"

Check the install

camcalib --help

Seven subcommands should be listed — calibrate, cam, cam-imu, apply-calib, roundtrip, run, analyze. They are documented in the CLI reference.

Next

Calibrate something in Quickstart.