Adaptive token division-based transformer for visual object tracking
Dan Tian, Dongxin Liu, Xiao Wang · 2026
Read Adaptive token division-based transformer for visual object tracking by Dan Tian, Dongxin Liu, Xiao Wang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0351112.txt.
Description
Dan Tian, Dongxin Liu, and Xiao Wang propose in 2026 a smarter way for video software to follow one object across frames. Standard systems compare every part of the search area with the target template, so look alike background items can confuse them. Their design splits image tokens in a learned way and masks attention so each group is handled fittingly.
Six public test sets were used to check whether the tracker stayed locked on target. Results supported the design, the authors report, with the splitting step helping the system tell object from background. The work matters for uses from filming to driving aids, where losing the target carries a cost.
Catalog record
| CREATOR | Dan Tian, Dongxin Liu, Xiao Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32918 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |