Research map / Detection and segmentation / Tracking
TR-MoE: Temporal Reliability-Aware Mixture-of-Experts for Robust Tracking
Tianle Wang, Xiangyang Yang, Jihua Zhu, Binrui Liu, Yanzhao Li, Shuiwang Li
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 |
|---|---|---|---|
| fuse SAM2-T with LoRAT-L378 via TR-MoE | LoRAT-L378 · LaSOT | AUC: 75.1 -> 77.3 | Improved |
| fuse SAM2-T with LoRAT-L378 via TR-MoE | SAM2-T · LaSOT | AUC: 67.3 -> 77.3 | Improved |
| TR-MoE on harder extension | LoRAT-L378 · LaSOText | AUC: 56.6 -> 58.4 | Improved |
| TR-MoE on GOT-10k | LoRAT-L378 · GOT-10k | AO: 77.5 -> 81.4 | Improved |
| TR-MoE on weak lightweight tracker | SGLATrack · LaSOT | AUC: 63.0 -> 70.5 | Improved |
| TR-MoE on ODTrack-B384, which has own temporal modeling | ODTrack-B384 · LaSOT | AUC: 73.2 -> 75.9 | Improved |
| TR-MoE on SGLATrack in GOT-10k | SGLATrack · GOT-10k | AO: 66.3 -> 78.6 vs SAM2 78.5 | No clear effect |
| replace MLP backbone by Mamba (ablation) | TR-MoE (Linear) · LaSOT | AUC: 72.54 -> 74.53 | Improved |
| Mamba instead of Transformer encoder (ablation) | TR-MoE (Linear) · LaSOT | AUC: 74.34 -> 74.53 | Improved |
| replace Linear with KAN embedding and head (ablation) | TR-MoE (Mamba) · LaSOT | AUC: 74.53 -> 75.22 | Improved |
| decouple head into direction and magnitude (ablation) | TR-MoE · LaSOT | AUC: 71.01 -> 72.48 | Improved |
| replace average fusion with adaptive fusion (ablation) | TR-MoE · LaSOT | AUC: 72.48 -> 74.82 | Improved |
Closest papers by meaning
- SENTRY: SAM2-Enhanced Neighbor-Aware and Temporally Reasoned Memory for Visual Tracking (ECCV 2026 · similarity 0.86)
- Learning Dynamic Siamese Network for Visual Object Tracking (ICCV 2017 · similarity 0.86)
- Delving into Dynamic Scene Cue-Consistency for Robust 3D Multi-Object Tracking (AAAI 2026 · similarity 0.85)
- ODTrack: Online Dense Temporal Token Learning for Visual Tracking (AAAI 2024 · similarity 0.85)
- When Trackers Fail: VLM-Guided Verification and Recovery for Robust Video Object Segmentation (NeurIPS 2026 · similarity 0.85)
- SPMTrack: Spatio-Temporal Parameter-Efficient Fine-Tuning with Mixture of Experts for Scalable Visual Tracking (CVPR 2025 · similarity 0.84)
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.
