rosbagkit

rosbagkit is a lightweight toolkit for reading ROS bagfiles and extracting common sensor streams such as RGB images, depth images, odometry, IMU data, GPS data, and TF transforms.

The repository is organized around:

• a Python package in src/rosbagkit
• extraction scripts in scripts/
• YAML-driven extraction examples in config/extract/

Requirements

• Python 3.12+
• ROS bagfiles supported by rosbags

Install

uv sync
source .venv/bin/activate

uv must be installed separately before running these commands.

Usage

Extract bag contents from a YAML config

Use one of the template configs in config/extract/ as a starting point and replace the placeholder paths, topics, and scene entries with your own data.

python scripts/extract_bagfile.py config/extract/UT-SARA-GQ.yaml

Each config follows this shape:

bagfile_root: /path/to/bagfiles
output_root: /path/to/output

topics:
  /camera/color/image_raw:
    format: image
    outdir: image_raw
  /gps/fix:
    format: csv
    outpath: gps.csv

scenes:
  example_scene:
    bagfiles:
      - example_run/example_0.bag
    start: 0
    end: -1

Supported topic output formats currently include:

• image
• depth
• pointcloud_depth
• csv

Optional stereo rectification can be configured directly in the YAML file:

rectification:
  enabled: true
  left_topic: /stereo/left/image_raw
  right_topic: /stereo/right/image_raw
  left_calib: /path/to/left_intrinsics.yaml
  right_calib: /path/to/right_intrinsics.yaml
  extrinsics: /path/to/left_to_right.yaml
  threshold: 0.005
  output_dir: 2d_rect
  left_subdir: cam_left
  right_subdir: cam_right
  timestamp_file: timestamps.txt

When enabled, the configured stereo topics are rectified in memory and only the final rectified images are written to disk. See config/extract/UT-SARA-GQ.yaml for a public GQ example.

Extract a TF transform from a bagfile

python scripts/extract_tf.py /path/to/example.bag --src base_link --tgt camera_link

This writes a YAML file next to the bagfile containing the resolved transform chain.