boxmot
BoxMOT: Pluggable Python and C++ SOTA multi-object tracking modules with support for axis-aligned and oriented bounding boxes
What is boxmot?
BoxMOT: Pluggable Python and C++ SOTA multi-object tracking modules with support for axis-aligned and oriented bounding boxes
Overview of boxmot
What it is
BoxMOT is a pluggable Python and C++ multi-object tracking module that supports both axis-aligned (AABB) and oriented bounding boxes (OBB). It provides independent detector, segmentor, appearance-encoder, and tracker components built around validated Torch structures. Pipelines compose these components; the CLI owns sources, outputs, materialized datasets, evaluation, tuning, research, and ReID workflows.
Key features
- One interface for
track,materialize,materialize --time-variant,eval,tune,research,train-reid,eval-reid,compare-reid,export,export-edgetam,install, and nativebuildworkflows. - Swappable components with explicit capabilities and requirements.
- Immutable, keyed Parquet builds with reusable detections, masks, and embeddings.
- Support for both AABB and OBB tracking paths.
- Optional production-ready native C++ tracker implementations with the same metrics as the Python path, opted into via
--tracker-backend cppand embeddable in standalone C++ projects via CMake. - A structured Python API for embedding components and pipelines in applications.
Use cases
BoxMOT is suitable for multi-object tracking in video streams, evaluation of tracking algorithms on standard benchmarks (e.g., MOT17, SportsMOT, MMOT), and research workflows involving ReID training and comparison. It can be used with any detection model and supports both 2D and 3D sensor fusion with ego-motion compensation via EagerMot.
Getting started
bashpip install boxmotboxmot --helpFor detailed installation instructions, including source checkouts and mode-specific extras, see the repository README.
License and pricing
BoxMOT is licensed under AGPL-3.0. It is an open-source project with no stated pricing.
Based on the project's public README and repository metadata.