Research map / Vision-language and multimodal / Multimodal LLMs
ATCTrack: Aligning Target-Context Cues with Dynamic Target States for Robust Vision-Language Tracking
Xiaokun Feng, Shiyu Hu, Xuchen Li, Dailing Zhang, Meiqi Wu, Jing Zhang, Xiaotang Chen, Kaiqi Huang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add textual target-context guidance (ablation) | ATCTrack-B naive · TNL2K | AUC: 65.5 -> 66.5 | Improved |
| add visual target-context guidance (ablation) | ATCTrack-B naive · TNL2K | AUC: 65.5 -> 66.7 | Improved |
| combine textual and visual target-context guidance (ablation) | ATCTrack-B naive · LaSOT | AUC: 72.5 -> 74.7 | Improved |
| add target words awareness to textual modeling (ablation) | naive text fusion · TNL2K | AUC: 65.5 -> 65.9 | Improved |
| add context words calibration (ablation) | ATCTrack-B with target words awareness · TNL2K | AUC: 65.9 -> 66.5 | Improved |
| use dual-type textual features (initial and calibrated) instead of calibrated only (ablation) | ATCTrack-B textual guidance · LaSOT | AUC: 73.2 -> 73.6 | Improved |
| use RoI local region of predicted box for visual context (ablation) | naive visual baseline · LaSOT | AUC: 72.5 -> 71.1 | Hurt |
| use search region with cropped local mask for visual context (ablation) | naive visual baseline · LaSOT | AUC: 72.5 -> 71.9 | Hurt |
| use search region with global mask instead of crop mask (ablation) | visual target-context module · LaSOT | AUC: 71.9 -> 73.7 | Improved |
| use HiViT backbone instead of ViT (ablation) | ATCTrack-B · TNL2K | AUC: 65.8 -> 67.5 | Improved |
| add dynamic template (ablation) | ATCTrack-B · LaSOT | AUC: 73.2 -> 74.7 | Improved |
| add target word supervision labels (BCE loss) (ablation) | ATCTrack-B · TNL2K | AUC: 67.0 -> 67.5 | Improved |
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.
- MemVLT — MemVLT: Vision-Language Tracking with Adaptive Memory-based Prompts (NeurIPS 2024) · better on 5 of 5 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
- Unifying Visual and Vision-Language Tracking via Contrastive Learning (AAAI 2024 · similarity 0.90)
- ChatTracker: Enhancing Visual Tracking Performance via Chatting with Multimodal Large Language Model (NeurIPS 2024 · similarity 0.88)
- MemVLT: Vision-Language Tracking with Adaptive Memory-based Prompts (NeurIPS 2024 · similarity 0.87)
- Less Is More: Token Context-Aware Learning for Object Tracking (AAAI 2025 · similarity 0.86)
- MIMTrack: In-Context Tracking via Masked Image Modeling (AAAI 2025 · similarity 0.86)
- SVLTA: Benchmarking Vision-Language Temporal Alignment via Synthetic Video Situation (CVPR 2025 · similarity 0.86)
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.
