AI Paper Atlas

Research map / Label-efficient and robust learning / Test-time adaptation

Dual-level Adaptation for Multi-Object Tracking: Building Test-Time Calibration from Experience and Intuition

Wen Guo, Pengfei Zhao, Zongmeng Wang, Yufan Hu, Junyu Gao

CVPR 2026 · Label-efficient and robust learning › Test-time adaptation · approach: Pseudo-label test-time 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 TCEI test-time calibration (Intuitive plus Experiential)MOTIP · DanceTrack testHOTA: 69.5 -> 70.6Improved
add TCEI test-time calibrationMOTIP · SportsMOT testHOTA: 72.6 -> 73.0Improved
add the Intuitive system (transient memory) (ablation)TCEI · DanceTrackHOTA: 69.5 -> 70.5Improved
add the Experiential system (experience cache) (ablation)TCEI · DanceTrackHOTA: 69.5 -> 70.4Improved
add the Experiential system on top of the Intuitive system (ablation)TCEI · DanceTrackHOTA: 70.5 -> 70.6Improved
use confident objects as temporal priors only (ablation)TCEI · DanceTrackHOTA: 69.5 -> 69.6No clear effect
use uncertain objects as reflective cases only (ablation)TCEI · DanceTrackHOTA: 69.5 -> 70.2Improved
combine confident and uncertain objects (ablation)TCEI · DanceTrackHOTA: 70.2 -> 70.5Improved
replace naive averaging with selective uncertainty-based calibration (ablation)TCEI · DanceTrackHOTA: 69.7 -> 70.6Improved
replace entropy-based selection with selective calibration (ablation)TCEI · DanceTrackHOTA: 69.9 -> 70.6Improved
apply Tent entropy-minimization backpropagationMOTIP · DanceTrack testHOTA: 69.5 -> 69.4No clear effect
apply Tent instead of TCEI on SportsMOTMOTIP · SportsMOT testHOTA: 72.6 -> 72.8No clear effect

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