DataScope Studio turns robotics, industrial, and sensor datasets into interactive Rerun visualizations. Import local data, review the automatically generated mapping, build a recording, and open it in Rerun without writing a custom visualization script.
Download DataScope Studio | Read the documentation
- Download the installer for your platform from GitHub Releases.
- Launch DataScope Studio. The installed app starts the local API, Python runtime, and Rerun integration automatically.
- On Dashboard, create a project or select an existing one.
- Choose a data source by selecting a file or folder. Use Copy into project when you want a self-contained local project, or Reference original path when the source data should stay in place.
- Select Import & Auto Map. DataScope inspects the source, recommends a template, and creates a draft mapping.
- In Import Workflow, check Schema Inspector and Mapping Editor. Adjust the time field, time unit, semantic type, entity path, or template if needed.
- Select Validate Mapping, then Confirm Mapping.
- Enter an output name and select Build .rrd + .rbl.
- Select Open in Rerun to inspect the generated recording.
The installer includes the desktop application, local API, Python runtime, DataScope packages, and Rerun. Python, Node.js, npm, and Rerun do not need to be installed separately for normal desktop use.
- Inspect heterogeneous data in one repeatable workflow.
- Automatically map timestamps, scalar values, states, logs, images, detections, and point clouds.
- Generate reusable Rerun
.rrdrecordings and.rblblueprints. - Organize projects, sources, mappings, recordings, jobs, tags, parameters, and query exports locally.
- Search common conditions such as errors, low battery, detection failures, topic summaries, time sync issues, and state duration.
- Run offline robot diagnostics and export JSON, CSV, or HTML reports.
- Export and reopen complete DataScope project packages.
- The Tauri desktop app talks directly to the local FastAPI server first and only falls back to the Tauri proxy when direct requests fail, reducing normal API overhead.
- API startup warms the workspace in the background, so health checks and window startup do not wait for a full workspace scan.
- Import & Auto Map uses one import workflow request for adding the source, inspecting it, selecting templates, saving a draft mapping, previewing rows, and validating the mapping.
- Conversion jobs throttle progress writes to SQLite so large conversions do not spend excessive time updating job metadata.
- Query templates prefer the lightweight query index created during conversion and stream rows with limits where possible to reduce memory pressure on large projects.
| Source | Examples | Typical Template |
|---|---|---|
| Tables | CSV, JSONL/NDJSON | Sensor Monitor |
| Text tables and logs | TSV, TXT, LOG, DAT, LST, LIST with comma, tab, semicolon, pipe, or whitespace delimiters | Sensor Monitor |
| Computer vision | Image files or folders with optional annotations.json and predictions.json sidecars |
CV Detection |
| Point clouds | PLY, PCD, NPY, NPZ, XYZ, XYZN, XYZRGB, PTS, ASC files or frame directories | Robotics Debug |
| Robotics recordings | MCAP and ROS2 DB3 bags, including split bag directories | Robotics Debug |
DataScope is local-first. Source data, mappings, recordings, and the SQLite catalog remain in your local workspace unless you explicitly export a project package.
Download the installer for your computer from GitHub Releases.
| System | Download |
|---|---|
| Windows 10/11 x64 | DataScope-Studio-v0.4.0-windows-x86_64-setup.exe |
| macOS Apple Silicon | DataScope-Studio-v0.4.0-macos-aarch64.dmg |
| macOS Intel | DataScope-Studio-v0.4.0-macos-x86_64.dmg |
| Debian/Ubuntu x64 | DataScope-Studio-v0.4.0-linux-amd64.deb |
| Other Linux x64 | DataScope-Studio-v0.4.0-linux-x86_64.AppImage |
The v0.4.0 packages are unsigned prerelease builds:
- Windows: if SmartScreen appears, choose More info and then Run anyway after confirming the installer came from this repository.
- macOS: drag DataScope Studio to Applications. On first launch, Control-click the app, choose Open, and confirm. You can also allow it under System Settings > Privacy & Security.
- Linux AppImage: make it executable before launching:
chmod +x DataScope-Studio-v0.4.0-linux-x86_64.AppImage
./DataScope-Studio-v0.4.0-linux-x86_64.AppImageFor the Debian package:
sudo apt install ./DataScope-Studio-v0.4.0-linux-amd64.deb- Dashboard: create projects, import a source, open project packages, and review recent recordings.
- Import Workflow: choose a recommended template, edit or save mapping templates, validate mappings, build Rerun artifacts, and open them in Rerun.
- Recordings & Queries: reopen recordings, add tags, run query templates, export query results, compare scalar metrics, and inspect background jobs.
- Diagnostics: run offline robot health reports across all recordings or selected recordings, tune thresholds, and export reports.
- Extensions & Settings: install plugins/templates, run batch imports, export project packages, set default export and artifact folders, and inspect application logs.
For a headerless CSV, select No header and enter ordered column names such as
timestamp,x,y,z,rx,ry,rz. For TXT/LOG/DAT/LST sources, DataScope auto-detects table
delimiters when possible and falls back to a line-number plus message log view when the
file is unstructured.
Projects are stored under ~/.datascope-studio by default. When no custom artifact
folder is set, recordings and blueprints are written to the project recordings/ and
blueprints/ folders. Set the Rerun artifact folder in Settings or the conversion
card to place the .rrd and .rbl files together somewhere else.
Use Export Project Package to create a portable .datascope.zip package. Reopen
existing packages from Open Package on the dashboard. Query and diagnostics exports
are written to the project exports/ folder unless you choose a specific output path.
The desktop app is the primary user interface. For scripted imports from a source
checkout or an environment where datascope-cli is installed:
datascope inspect tests/fixtures/sample_sensor.csv
datascope import tests/fixtures/sample_sensor.csv --project demo --out run_001
datascope import /path/to/run.mcap --project robot_demo --template robotics_debug --out robot_run
datascope recordings --project demo
datascope query --project demo --template low_battery --threshold 0.2
datascope diagnose --project robot_demo --preset strict --format html --out robot_diagnostics.html
datascope project export --project demo --out demo.datascope.zipFrom a source checkout, create a development environment first:
python3 -m venv .venv
. .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -r requirements-dev.txtThen start the development desktop app:
cd apps/desktop
npm install
npm run tauri:dev- Installation and packaging
- Import and conversion workflow
- Troubleshooting
- Developer setup
- Architecture and API documentation
DataScope Studio is licensed under the Apache License 2.0.