MGIT Leaderboard (OPE Mechanism with Tiny Version, Story Granularity)

Up-to-date multi-modal global instance tracking performance of baseline and submitted results on MGIT. All entries are ranked based on their N-PRE scores with traditional OPE mechanism.

Up-to-Date Leaderboard

Instructions:

Currently we have:

Rank Method N-PRE PRE SRIoU SRDIoU SRGIoU Hz Hardware Language Date(UTC) Author
1 JointNLT 0.773 1 0.433 2 0.600 2 0.588 2 0.589 2 7.85 Titan RTX python 2024-01-21, 01:31:48 Official
2 MemVLT 0.738 2 0.560 1 0.631 1 0.623 1 0.625 1 16.04 3 3090 python 2024-05-03, 11:38:58 Anonymous Author
3 VLT_TT 0.616 3 0.322 3 0.480 3 0.468 3 0.470 3 7.58 Titan RTX python 2024-01-21, 02:02:48 Official
4 VLT_SCAR 0.403 0.127 0.184 0.171 0.179 19.02 1 Titan RTX python 2024-01-21, 02:01:23 Official
5 SNLT 0.230 0.005 0.040 0.026 0.028 18.60 2 Titan RTX python 2024-01-21, 02:04:19 Official