DEV Community

Vraj Patel
Vraj Patel

Posted on

OpenTrace ML: Help Build Privacy-Aware Road Intelligence in Python

#machinelearning

OpenTrace ML is an Apache-2.0 Python library for experimenting with road intelligence using computer-vision detections, incremental traffic forecasts, GPS traces, and map signals.

The current Stage 4 pre-alpha version can:

  • parse RDD2022-style road-damage annotations;
  • adapt model-independent callable detectors;
  • calculate detection and forecasting metrics;
  • learn traffic patterns incrementally;
  • place detections on GPS traces and export GeoJSON;
  • prepare consented GPX traces with pseudonymous trip IDs;
  • validate map-matching results without depending on one routing engine;
  • calculate transparent route-reliability scores.

Try it locally 

git clone https://github.com/vrajpatell/opentrace-ml.git
cd opentrace-ml
python -m venv .venv
source .venv/bin/activate
pip install -e '.[dev]'

python examples/road_damage_route_demo.py tests/fixtures/rdd_sample.xml

OPENTRACE_PSEUDONYM_KEY='replace-with-a-secret' \
  python examples/map_match_fixture.py

python -m pytest -q
Enter fullscreen mode Exit fullscreen mode

These examples use tiny original fixtures and require no external dataset download or routing service.

Where contributors can help

Our highest-priority task is creating a tiny offline OpenStreetMap integration fixture.

Other open tasks cover:

  • per-class road-damage metrics;
  • an optional MMDetection/RTMDet adapter;
  • GPS recording-gap handling;
  • privacy-preserving aggregation thresholds;
  • pseudonym-key rotation and retention guidance.

Contributions involving Python, machine learning, computer vision, GIS, routing, privacy, testing, and documentation are welcome.

Repository: https://github.com/vrajpatell/opentrace-ml

Disclosure: This announcement was drafted with AI assistance.

Top comments (0)