Research map / Vision-language and multimodal / Human and embodied modeling
Joint Visual Grounding and Tracking With Natural Language Specification
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace separate relation encoders with multi-source relation modeling (ablation) | JointNLT · LaSOT | AUC: 0.518 -> 0.536 | Helped |
| Replace separate relation encoders with multi-source relation modeling (ablation) | JointNLT · TNL2K | AUC: 0.491 -> 0.511 | Helped |
| Add a target decoder (ablation) | JointNLT MSRM · LaSOT | AUC: 0.536 -> 0.549 | Helped |
| Add a target decoder (ablation) | JointNLT MSRM · TNL2K | AUC: 0.511 -> 0.524 | Helped |
| Add temporal modeling from historical target patches (ablation) | JointNLT with target decoder · LaSOT | AUC: 0.549 -> 0.561 | Helped |
| Add temporal modeling from historical target patches (ablation) | JointNLT with target decoder · TNL2K | AUC: 0.524 -> 0.541 | Helped |
| Guide temporal modeling with the NL CLS token (ablation) | JointNLT SGTM · LaSOT | Precision: 0.581 -> 0.593 | Helped |
| Guide temporal modeling with the NL CLS token (ablation) | JointNLT SGTM · TNL2K | Precision: 0.540 -> 0.550 | Helped |
| Use JointNLT rather than CTRNLT for language-only tracking | Natural-language-initialized tracker · LaSOT | AUC: 0.520 -> 0.569 | Helped |
| Use JointNLT rather than VLTTT for language-and-box tracking | Tracker initialized with NL and BB · TNL2K | AUC: 0.531 -> 0.569 | Helped |
| Use JointNLT rather than VLTTT for language-and-box tracking | Tracker initialized with NL and BB · OTB99 | AUC: 0.764 -> 0.653 | Hurt |
| Use JointNLT rather than VLTVG for visual grounding | Visual grounding model · RefCOCOg-google val | Top-1 accuracy: 0.7298 -> 0.7007 | Hurt |
Closest papers by meaning
- Look Around Before Locating: Considering Content and Structure Information for Visual Grounding (AAAI 2025 · similarity 0.89)
- Your Large Vision-Language Model Only Needs A Few Attention Heads For Visual Grounding (CVPR 2025 · similarity 0.89)
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.89)
- Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning (AAAI 2026 · similarity 0.88)
- GroundingME: Exposing the Visual Grounding Gap in MLLMs through Multi-Dimensional Evaluation (CVPR 2026 · similarity 0.88)
- GroundVLP: Harnessing Zero-Shot Visual Grounding from Vision-Language Pre-training and Open-Vocabulary Object Detection (AAAI 2024 · similarity 0.88)
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.