UniBench300

September 22, 2025 ยท View on GitHub

Limited by the file size, the whole UniBench300 can not be shared here. Thus only sevral samples are displayed here.

image

UniBench300 contains RGBT, RGBD, and RGBE data simultaneously.

Benchmark-Toolkit

Download the UniBench300* files to maintain the toolkit for UniBench300.

UniBench_Toolkit.zip is the main file and create 'tracking_results' after unziped. Then unzip other files into this dictory.

image

[Toolkit]: URL CODE:TZYD [Dataset]: URL CODE:TZYD

Results

To make this work more convincing, the result files of original, mixed, and CL-boosted versions on LasHeR,
DepthTrack, and VisEvent are provided. (SymTrack*)

Code

StageInitialisedInvolved DataSaved Model
1RandomVisEventSymTrack-e.pth.tar
2SymTrack-e.pth.tarVisEvent, LasHeRSymTrack-et.pth.tar
3SymTrack-et.pth.tarVisEvent, LasHeR, DepthTrackSymTrack-etd.pth.tar
Since the proposed continual unification process mainly differs in the trainng stage,
the core training file for configuration is provided (base-t.yaml)