AI Paper Atlas

Research map / Detection and segmentation / Tracking

TCTrack: Temporal Contexts for Aerial Tracking

Ziang Cao, Ziyuan Huang, Liang Pan, Shiwei Zhang, Ziwei Liu, Changhong Fu

CVPR 2022 · Detection and segmentation › Tracking · approach: Transformer-based video object tracking · 269 citations (OpenAlex)

Check 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 temporal information filtering to the temporal transformer (ablation)AT-Trans · UAV123success: 0.550 -> 0.573Improved
Add temporal information filtering to the temporal transformer (ablation)AT-Trans · UAV123precision: 0.750 -> 0.765Improved
Use multi-frame rather than single-frame training (ablation)AT-Trans without TIF, convolutional initialization · UAV123success: 0.550 -> 0.508Hurt
Use convolutional rather than random temporal-prior initialization (ablation)AT-Trans with TIF and multi-frame training · UAV123success: 0.586 -> 0.604Improved
Use the previous temporal prior rather than the current similarity map as encoder query (ablation)AT-Trans with TIF, multi-frame training, and convolutional initialization · UAV123success: 0.580 -> 0.604Improved
Add temporal context filtering and the full AT-Trans configuration rather than using the unfiltered Transformer (ablation)AT-Trans · UAV123success: 0.550 -> 0.604Improved
Increase TAdaConv sequence length from 1 to 2 (ablation)TAdaConv · UAV123success: 0.561 -> 0.573Improved
Increase TAdaConv sequence length from 2 to 3 (ablation)TAdaConv · UAV123success: 0.573 -> 0.580Improved
Increase TAdaConv sequence length from 1 to 3 (ablation)TAdaConv · UAV123precision: 0.749 -> 0.776Improved
Use the full TCTrack framework rather than the SiamRPN++ baselineUAVTrack112_L · UAVTrack112_Lsuccess: 0.559 -> 0.582Improved
Use the full TCTrack framework rather than the SiamRPN++ baselineUAVTrack112_L · UAVTrack112_Lprecision: 0.773 -> 0.786Improved
Add temporal context to feature extraction and similarity-map refinement (ablation)TCTrack compared with Transformer · UAV123success: 9.8%Improved

Later papers that compared with this method

2 later accepted papers report TCTrack as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.

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