Skip to content

Quickstart

Quickstart

Install BoxMOT and inspect the CLI:

pip install boxmot
boxmot --help

Track a video:

boxmot track --detector yolov8n --reid osnet_x0_25_msmt17 --tracker botsort --source video.mp4 --save

Benchmark a tracker on a built-in config:

boxmot eval --benchmark mot17 --split ablation --tracker boosttrack --verbose

Research tracker code changes on a built-in config:

boxmot research --benchmark mot17 --split ablation --tracker bytetrack --proposal-model openai/gpt-5.4 --max-metric-calls 24

Use the high-level Python API:

from boxmot import Boxmot

boxmot = Boxmot(detector="yolov8n", reid="lmbn_n_duke", tracker="boosttrack")
run = boxmot.track(source="video.mp4", save=True)
print(run)

metrics = boxmot.val(benchmark="mot17-mini")
print(metrics)

The high-level Python API is available directly from boxmot. Shared CLI and Python defaults still come from boxmot/configs/modes.yaml so detector, ReID, tracker, and runtime defaults stay aligned across both entry points.

Next steps: