AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

SFDATrack: Generalized Source-Free Domain Adaptive Tracking Under Adverse Weather Conditions

Siyuan Yao, Ziqi Wang, Junqi Huang, Ruiqi Yu, Wenqi Ren, Xiaochun Cao

ECCV 2026 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Geometric alignment and synthetic-to-real adaptation · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add DIM module (ablation)SFDATrack · NAT2021AUC: 51.22 -> 53.56Improved
add HPP module (ablation)SFDATrack · NAT2021AUC: 51.22 -> 53.24Improved
add both DIM and HPP (ablation)SFDATrack · NAT2021AUC: 51.22 -> 56.78Improved
add domain-specific alignment (ablation)SFDATrack · UAVDark70AUC: 58.29 -> 59.65Improved
add domain-invariant alignment (ablation)SFDATrack · UAVDark70AUC: 58.29 -> 59.55Improved
add both DSA and DIA (ablation)SFDATrack · UAVDark70AUC: 58.29 -> 61.22Improved
replace Transformer with Uni-Mamba in DIM (ablation)SFDATrack · UAVDark70AUC: 56.77 -> 58.85Improved
make Mamba bidirectional (Bi-Mamba) (ablation)SFDATrack · UAVDark70AUC: 58.85 -> 60.33Improved
use full DIM instead of Bi-Mamba (ablation)SFDATrack · UAVDark70AUC: 60.33 -> 61.22Improved
update EMA each epoch instead of each batch (ablation)SFDATrack · NAT2021AUC: 55.20 -> 56.78Improved
update EMA each epoch instead of every 5 epochs (ablation)SFDATrack · NAT2021AUC: 54.32 -> 56.78Improved
balanced DSA/DIA weights 0.5/0.5 instead of 0.1/0.5 (ablation)SFDATrack · NAT2021AUC: 55.09 -> 56.78Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy