Skip to content

Very first runnable BoT-SORT. Implement camera motion compensation wi…#232

Draft
tstanczyk95 wants to merge 4 commits intoroboflow:developfrom
tstanczyk95:botsort_dev
Draft

Very first runnable BoT-SORT. Implement camera motion compensation wi…#232
tstanczyk95 wants to merge 4 commits intoroboflow:developfrom
tstanczyk95:botsort_dev

Conversation

@tstanczyk95
Copy link
Collaborator

@tstanczyk95 tstanczyk95 commented Feb 6, 2026

…th orb.

What does this PR do?

Related Issue(s):

Type of Change

  • New feature (non-breaking change that adds functionality)

Testing

  • I have tested this change locally
  • I have added/updated tests for this change

Test details:
Running the new algorithm (BoT-SORT) implementation on a full test sets of MOT17, SportsMOT and SoccerNet-tracking (no crashing, no errors) and obtaining the performance results from official evaluation server.

Checklist

  • My code follows the style guidelines of this project
  • I have performed a self-review of my own code
  • I have commented my code where necessary, particularly in hard-to-understand areas
  • My changes generate no new warnings or errors
  • I have updated the documentation accordingly (if applicable)

Additional Context

It already performs better than the previously implemented algorithms (e.g. ByteTrack, the baseline). Development is still in progress. Currently two camera motion compensation (CMC) mechanisms are implemented: ORB, SIFT, sparse optical flow.

Initiating it as:
tracker = BoTSORTTracker(enable_cmc=True, cmc_method="sift")

Mind that its update function requires current frame:
dets = tracker.update(dets, frame)

Note: SIFT-bsaed CMC is considerably and noticeably slower than the ORB and sparse optical flow ones.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant