Research map / Detection and segmentation / Tracking
RT-RMOT: A Dataset and Framework for RGB-Thermal Referring Multi-Object Tracking
Yanqiu Yu, Zhifan Jin, Sijia Chen, Tongfei Chu, En Yu, Liman Liu, Wenbing Tao
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 GSPO-based RL fine-tuning to Qwen2.5-VL-3B with RGB input (ablation) | RTrack · RefRT test (RGB input) | HOTA: 2.09 -> 12.49 | Improved |
| add GSPO-based RL fine-tuning to Qwen2.5-VL-3B with RGBT input (ablation) | RTrack · RefRT test (RGBT input) | HOTA: 4.98 -> 15.53 | Improved |
| add thermal input to RGB input after RL fine-tuning (ablation) | RTrack · RefRT test | HOTA: 12.49 -> 15.53 | Improved |
| add thermal input to RGB input, zero-shot (ablation) | Qwen2.5-VL-3B · RefRT test | HOTA: 2.09 -> 4.98 | Improved |
| add Clipped Advantage Scaling (ablation) | RTrack · RefRT test (RGBT input) | HOTA: 5.27 -> 15.53 | Improved |
| add Structured Output Reward (ablation) | RTrack · RefRT test (RGBT input) | HOTA: 11.24 -> 15.53 | Improved |
| add Comprehensive Detection Reward (ablation) | RTrack · RefRT test (RGBT input) | HOTA: 11.16 -> 15.53 | Improved |
| add Comprehensive Detection Reward (ablation) | RTrack · RefRT test (RGBT input) | DetRe: 28.44 -> 20.15 | Hurt |
| add Clipped Advantage Scaling (ablation) | RTrack · RefRT test (RGBT input) | DetPr: 28.11 -> 22.78 | Hurt |
| use Qwen2.5-VL-3B instead of LLaVA-NeXT-8B in framework, zero-shot (ablation) | RTrack · RefRT test (RGBT input), zero-shot | HOTA: 2.99 -> 4.98 | Improved |
| use Qwen2.5-VL-3B instead of InternVL2-8B, zero-shot (ablation) | RTrack · RefRT test (RGBT input), zero-shot | HOTA: 0.82 -> 4.98 | Improved |
| use Qwen2.5-VL-3B instead of LLaVA-7B, zero-shot (ablation) | RTrack · RefRT test (RGBT input), zero-shot | HOTA: 1.13 -> 4.98 | Improved |
Closest papers by meaning
- Collaborating Vision, Depth, and Thermal Signals for Multi-Modal Tracking: Dataset and Algorithm (NeurIPS 2025 · similarity 0.88)
- Single-Model and Any-Modality for Video Object Tracking (CVPR 2024 · similarity 0.88)
- SUTrack: Towards Simple and Unified Single Object Tracking (AAAI 2025 · similarity 0.88)
- Temporal Adaptive RGBT Tracking with Modality Prompt (AAAI 2024 · similarity 0.87)
- SDSTrack: Self-Distillation Symmetric Adapter Learning for Multi-Modal Visual Object Tracking (CVPR 2024 · similarity 0.86)
- Cross-View Referring Multi-Object Tracking (AAAI 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.
