Research map / Vision-language and multimodal / Cross-modal learning
Unified Multimodal Visual Tracking with Dual Mixture-of-Experts
Lingyi Hong, Jinglun Li, Xinyu Zhou, Kaixun Jiang, Pinxue Guo, Zhaoyu Chen, Runze Li, Xingdong Sheng, Wenqiang Zhang
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 Meta Merger to the baseline tracker (ablation) | OneTrackerV2 baseline (HiViT-B) · LasHeR | AUC: 50.4 -> 55.7 | Improved |
| add Meta Merger to the baseline tracker (ablation) | OneTrackerV2 baseline (HiViT-B) · DepthTrack | F-score: 51.7 -> 62.1 | Improved |
| add stochastic modality perturbation on top of Meta Merger (ablation) | OneTrackerV2 (HiViT-B) · DepthTrack missing-modality | F-score: 51.8 -> 53.2 | Improved |
| add stochastic modality perturbation on top of Meta Merger (ablation) | OneTrackerV2 (HiViT-B) · LaSOT | AUC: 71.3 -> 71.0 | No clear effect |
| add a single MoE (ablation) | OneTrackerV2 (HiViT-B) · Average of five benchmarks | Average: 62.0 -> 62.4 | Improved |
| replace single MoE with naive dual MoE (no decoupling or clustering) (ablation) | OneTrackerV2 (HiViT-B) · Average of five benchmarks | Average: 62.4 -> 62.8 | Improved |
| add Expert Decoupling loss to dual MoE (ablation) | OneTrackerV2 (HiViT-B) · DepthTrack missing-modality | F-score: 54.1 -> 54.7 | Improved |
| add Router Cluster loss to dual MoE (ablation) | OneTrackerV2 (HiViT-B) · Average of five benchmarks | Average: 62.8 -> 63.5 | Improved |
| add Router Cluster loss to dual MoE (ablation) | OneTrackerV2 (HiViT-B) · LasHeR missing-modality | AUC: 50.2 -> 52.7 | Improved |
| add dual MoE (full) cost (ablation) | OneTrackerV2 (HiViT-B) · Inference speed | FPS: 90 -> 72 | Hurt |
| increase expert projection rank beyond 16 (ablation) | DMoE · tracking accuracy | accuracy | Hurt |
| increase number of experts (ablation) | DMoE · tracking accuracy | accuracy | Mixed |
Closest papers by meaning
- Single-Model and Any-Modality for Video Object Tracking (CVPR 2024 · similarity 0.93)
- OneTracker: Unifying Visual Object Tracking with Foundation Models and Efficient Tuning (CVPR 2024 · similarity 0.90)
- Collaborating Vision, Depth, and Thermal Signals for Multi-Modal Tracking: Dataset and Algorithm (NeurIPS 2025 · similarity 0.89)
- Bi-directional Adapter for Multimodal Tracking (AAAI 2024 · similarity 0.89)
- Exploiting Multimodal Spatial-temporal Patterns for Video Object Tracking (AAAI 2025 · similarity 0.88)
- SDSTrack: Self-Distillation Symmetric Adapter Learning for Multi-Modal Visual Object Tracking (CVPR 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.
